WO2022009904A1 - Engine abnormality diagnosis method, engine abnormality diagnosis program, and engine abnormality diagnosis system - Google Patents

Engine abnormality diagnosis method, engine abnormality diagnosis program, and engine abnormality diagnosis system Download PDF

Info

Publication number
WO2022009904A1
WO2022009904A1 PCT/JP2021/025526 JP2021025526W WO2022009904A1 WO 2022009904 A1 WO2022009904 A1 WO 2022009904A1 JP 2021025526 W JP2021025526 W JP 2021025526W WO 2022009904 A1 WO2022009904 A1 WO 2022009904A1
Authority
WO
WIPO (PCT)
Prior art keywords
engine
abnormality
state quantity
abnormality diagnosis
model
Prior art date
Application number
PCT/JP2021/025526
Other languages
French (fr)
Japanese (ja)
Inventor
オレクシー ボンダレンコ
哲吾 福田
修二郎 宮川
健 宮地
Original Assignee
国立研究開発法人 海上・港湾・航空技術研究所
株式会社三井Ｅ＆Ｓマシナリー
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 国立研究開発法人 海上・港湾・航空技術研究所, 株式会社三井Ｅ＆Ｓマシナリー filed Critical 国立研究開発法人 海上・港湾・航空技術研究所
Priority to CN202180042708.1A priority Critical patent/CN115702288A/en
Priority to JP2022535361A priority patent/JP7450238B2/en
Priority to KR1020227043363A priority patent/KR20230009484A/en
Publication of WO2022009904A1 publication Critical patent/WO2022009904A1/en

Links

Images

Classifications

    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F02—COMBUSTION ENGINES; HOT-GAS OR COMBUSTION-PRODUCT ENGINE PLANTS
    • F02D—CONTROLLING COMBUSTION ENGINES
    • F02D41/00—Electrical control of supply of combustible mixture or its constituents
    • F02D41/22—Safety or indicating devices for abnormal conditions
    • F02D41/221—Safety or indicating devices for abnormal conditions relating to the failure of actuators or electrically driven elements
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F02—COMBUSTION ENGINES; HOT-GAS OR COMBUSTION-PRODUCT ENGINE PLANTS
    • F02D—CONTROLLING COMBUSTION ENGINES
    • F02D45/00—Electrical control not provided for in groups F02D41/00 - F02D43/00
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00—Testing or monitoring of control systems or parts thereof
    • G05B23/02—Electric testing or monitoring
    • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
    • G05B23/0243—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
    • G05B23/0254—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model based on a quantitative model, e.g. mathematical relationships between inputs and outputs; functions: observer, Kalman filter, residual calculation, Neural Networks
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F02—COMBUSTION ENGINES; HOT-GAS OR COMBUSTION-PRODUCT ENGINE PLANTS
    • F02D—CONTROLLING COMBUSTION ENGINES
    • F02D41/00—Electrical control of supply of combustible mixture or its constituents
    • F02D41/02—Circuit arrangements for generating control signals
    • F02D41/14—Introducing closed-loop corrections
    • F02D41/1401—Introducing closed-loop corrections characterised by the control or regulation method
    • F02D2041/1413—Controller structures or design
    • F02D2041/1415—Controller structures or design using a state feedback or a state space representation
    • F02D2041/1417—Kalman filter
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F02—COMBUSTION ENGINES; HOT-GAS OR COMBUSTION-PRODUCT ENGINE PLANTS
    • F02D—CONTROLLING COMBUSTION ENGINES
    • F02D41/00—Electrical control of supply of combustible mixture or its constituents
    • F02D41/02—Circuit arrangements for generating control signals
    • F02D41/14—Introducing closed-loop corrections
    • F02D41/1401—Introducing closed-loop corrections characterised by the control or regulation method
    • F02D2041/1433—Introducing closed-loop corrections characterised by the control or regulation method using a model or simulation of the system
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Definitions

  • the present invention relates to an engine abnormality diagnosis method, an engine abnormality diagnosis program, and an engine abnormality diagnosis system for diagnosing an engine abnormality using a mathematical engine model.
  • Patent Document 1 uses a constant gain extended Kalman filter (CGEKF) consisting of a Kalman filter gain optimally designed at a certain operating point of the engine and a nonlinear dynamic simulation model that faithfully models the actual engine. Then, if there is a difference between the observed variable and the observed variable estimate, the nonlinear dynamic simulation model can be adjusted by adjusting the tuning parameters that form part of the state variable of the model so that the difference is minimized.
  • CGEKF constant gain extended Kalman filter
  • paragraph 0010 describes an engine performance estimation system and method that constantly estimates, detects, and monitors performance deterioration due to aging and damage due to engine use by using a nonlinear dynamic simulation model of the engine and an estimation filter.
  • Patent Document 2 is a system for determining the flight parameters of an aircraft in real time during the flight of the aircraft, and the interdependence between at least two preselected flight parameters of the aircraft to be estimated.
  • a system comprising an extended Kalman filter configured on the basis of a flight dynamics equation to provide a congruent estimate of selected flight parameters during flight of an aircraft.
  • Patent Document 3 slow or intermittent while providing the controller with new process variable estimates during each execution cycle of the controller to allow the controller to generate control signals used to control the process.
  • a Kalman filter configured to generate an estimation of a process variable value from a different process feedback signal is disclosed.
  • Japanese Unexamined Patent Publication No. 2009-68359 Japanese Unexamined Patent Publication No. 2013-49408 Japanese Unexamined Patent Publication No. 2018-88289
  • Patent Document 1 provides a high-precision nonlinear dynamic simulation model that sufficiently reflects the effects of aging deterioration, damage, etc. that change the correlation between measurable items and non-measurable items.
  • the non-linear dynamic simulation model is sequentially tuned, the unmeasurable items of the engine are estimated from the measurable items, and the gas turbine engine is controlled using the estimation results. Therefore, it does not detect an abnormality in the engine at an early stage and diagnose its cause.
  • Patent Document 2 simultaneously estimates at least two flight parameters associated with each other by flight dynamics equations by an extended Kalman filter and uses them for flight control of an aircraft, and early detection of engine abnormality and diagnosis of the cause thereof. It's not something to do.
  • Patent Document 3 relates to an improvement of a Kalman filter in a process control system such as a chemical process or a petroleum process, and does not perform early detection of an abnormality in an engine and diagnosis of the cause thereof. Therefore, an object of the present invention is to provide an engine abnormality diagnosis method, an engine abnormality diagnosis program, and an engine abnormality diagnosis system capable of early detection of an engine abnormality and diagnosis of the cause thereof.
  • the engine abnormality diagnosis method is a program for diagnosing an engine abnormality using a mathematical engine model, and is an initial state quantity acquisition step for acquiring the initial state quantity of the engine model.
  • the engine model utilization step that applies the initial state quantity to the engine model and utilizes the engine model
  • the engine state estimation step that calculates the engine state based on the initial state quantity in the engine model and obtains the estimated state quantity
  • the engine The Kalman filtering step for acquiring the measured state quantity, the Kalman filtering step for applying the residual between the acquired measured state quantity and the calculated estimated state quantity to the nonlinear Kalman filter, and the Kalman gain obtained by applying the nonlinear Kalman filter to the engine model.
  • a factor analysis step that performs factor analysis on the correlation of state functions or residuals to obtain factor loadings
  • an abnormality detection step that calculates factor scores from factor loadings and detects abnormalities
  • factor loadings are applied to machine learning. It is characterized by executing a machine learning application step, an abnormality diagnosis step for diagnosing an abnormality based on machine learning, and an output step for outputting abnormality information including an engine abnormality diagnosis result.
  • an abnormality of the engine is performed based on the calculated factor score by performing factor analysis using the measured state quantity or the residual between the measured state quantity and the calculated estimated state quantity. Can be detected at an early stage.
  • the factor load can be applied to machine learning to diagnose the cause of engine abnormality.
  • the present invention according to claim 2 further executes a model update step for updating an engine model by applying the result obtained in the Kalman filtering step or the result of processing the acquired measurement state quantity to the engine model utilization step. It is characterized by that.
  • the engine model can be updated to keep the calculation accuracy of the estimated state quantity always high.
  • the present invention according to claim 3 is characterized in that the calculation of the correlation of the measured state quantity or the residual in the correlation calculation step is performed based on the correlation matrix. According to the third aspect of the present invention, the calculation accuracy of the measured state quantity or the correlation of the residual can be improved by a simple calculation.
  • the present invention according to claim 4 is characterized in that, in the factor analysis step, a factor loading is derived by performing singular value decomposition (SVD) based on a covariance matrix as a correlation matrix.
  • SVD singular value decomposition
  • the present invention according to claim 5 is characterized in that machine learning uses a self-organizing map (SOM).
  • SOM self-organizing map
  • the cause of an engine abnormality can be classified by using a self-organizing map (SOM) which is unsupervised machine learning.
  • the initial state quantity acquired in the initial state quantity acquisition step characterized in that it is a fuel supply amount including load (Q p) fuel pump rack position (h p) of the engine .
  • the estimated state You can get the quantity.
  • the present invention according to claim 7 is characterized in that the measured state quantity acquired in the measured state quantity acquisition step is the engine speed ( ne ).
  • the engine speed ( ne ) which is often measured as an engine, can be used for engine abnormality diagnosis, and the measurement accuracy can be improved.
  • the present invention of claim 8 wherein, as the estimated state quantity of the engine state estimating step, the supercharger rotational speed (n tc), scavenging pressure (P s), the scavenging temperature (T s), and the exhaust gas temperature (T e ) Is obtained.
  • useful supercharger speed in estimating the state of the engine (n tc), scavenging pressure (P s), the scavenging temperature (T s), or the exhaust gas temperature (T e) diagnostics for can be obtained.
  • the engine abnormality diagnosis program corresponding to claim 9 is a program for diagnosing an engine abnormality using a mathematical engine model, and is a step of acquiring an initial state amount in an engine abnormality diagnosis method using a computer. , Engine model utilization step, engine state estimation step, measurement state quantity acquisition step, Kalman filtering step, repeat step, correlation calculation step, factor analysis step, abnormality detection step, machine learning application step, abnormality diagnosis step, and output step. It is characterized by being executed. According to the invention of claim 9, an abnormality of the engine is detected based on the calculated factor score by performing factor analysis using the measured state quantity or the residual between the measured state quantity and the calculated estimated state quantity. It can be detected early. In addition, the factor load can be applied to machine learning to diagnose the cause of engine abnormality.
  • the engine In the engine abnormality diagnosis system corresponding to the tenth aspect, the engine, the condition input means for inputting the initial state amount of the engine model, the state amount measuring means for measuring the state of the engine and obtaining the measured state amount, and the engine.
  • the present invention is characterized by comprising a computer for executing the abnormality diagnosis method of the engine or an abnormality diagnosis program of the engine, and an information providing means for providing abnormality information including the diagnosis result of the abnormality of the engine output from the computer. According to the tenth aspect of the present invention, it is possible to use a computer to provide abnormality information including a result of early detection of an abnormality of an engine and diagnosis of the cause thereof.
  • the present invention according to claim 11 is characterized in that the engine model is updated by a computer. According to the eleventh aspect of the present invention, it is possible to easily improve the calculation accuracy of the estimated state quantity by updating the engine model with a computer.
  • the present invention according to claim 12 is characterized in that a state quantity measuring means obtains an engine speed (ne) as a measured state quantity.
  • the engine speed ( ne ) which is often measured as an engine, can be used for engine abnormality diagnosis, and the measurement accuracy can be improved.
  • the present invention is an information providing means, and as a result of diagnosing an abnormality, an engine supercharger rotation speed ( ntc ), a scavenging pressure (P s ), a scavenging temperature (T s ), and an exhaust gas temperature. and providing at least one result of the (T e).
  • useful supercharger speed in estimating the state of the engine (n tc), scavenging pressure (P s), the scavenging temperature (T s), or the exhaust gas temperature (T e) diagnostics for can be obtained.
  • the present invention according to claim 14 is characterized in that an abnormality control means for controlling an engine at the time of abnormality is provided based on the output of abnormality information. According to the 14th aspect of the present invention, by controlling the engine at the time of abnormality, it is possible to prevent the abnormality of the engine from getting worse or leading to a failure.
  • the present invention according to claim 15 is characterized in that anomalous information is provided by using a human interface means as an information providing means. According to the fifteenth aspect of the present invention, by providing the abnormality information of the engine from the human interface means, the crew and the like can quickly and appropriately respond to the abnormality.
  • the present invention according to claim 16 is characterized by comprising a transmission means for transmitting anomalous information provided by the information providing means to another place. According to the sixteenth aspect of the present invention, it is possible to know the diagnosis result including the abnormality of the engine in real time even at a place away from the engine.
  • the present invention according to claim 17 is characterized by comprising a state quantity measuring means, a connecting means for connecting a computer and an information providing means online. According to the 17th aspect of the present invention, a diagnostic result including an abnormality of an engine can be transmitted and provided online in real time.
  • an engine abnormality is performed based on the calculated factor score by performing factor analysis using the measured state quantity or the residual between the measured state quantity and the calculated estimated state quantity. Can be detected at an early stage.
  • the factor load can be applied to machine learning to diagnose the cause of engine abnormality.
  • the engine model is updated. Therefore, the calculation accuracy of the estimated state quantity can always be kept high.
  • the measurement state quantity or the residual correlation calculation accuracy can be improved by a simple calculation.
  • the factor analysis step when the factor loading is derived by singular value decomposition (SVD) based on the covariance matrix as the correlation matrix, it is essentially important regardless of the shape of the covariance matrix. Can be extracted and an abnormality in the engine can be detected at an early stage.
  • SVD singular value decomposition
  • the initial state quantity acquired in the initial state quantity acquisition step if a fuel supply amount including load (Q p) fuel pump rack position (h p) of the engine, important for estimating the state of the engine it is possible to obtain an estimated state quantity based on the fuel supply amount including load (Q p) fuel pump rack position (h p) of the engine is.
  • the measurement state quantity acquired by measuring the state quantity obtaining step when a rotational speed of the engine (n e), using the rotation speed of the large engines opportunity to measure the engine (n e) in the abnormality diagnosis of the engine It is possible to improve the measurement accuracy.
  • the engine is based on the calculated factor score by performing factor analysis using the measured state quantity or the residual between the measured state quantity and the calculated estimated state quantity. Abnormality can be detected at an early stage.
  • the factor load can be applied to machine learning to diagnose the cause of engine abnormality.
  • the engine abnormality diagnosis system of the present invention it is possible to provide abnormality information including the result of early detection of engine abnormality and diagnosis of the cause thereof by using a computer.
  • a state quantity measuring means In a state quantity measuring means, the case of obtaining a rotational speed of the engine (n e) as the measured state variables can be used in many engines opportunity to measure the engine rotational speed (n e) in the abnormality diagnosis of the engine , The measurement accuracy can also be improved.
  • the information providing unit as a diagnosis result of the abnormality, the supercharger rotation speed of the engine (n tc), scavenging pressure (P s), the scavenging temperature (T s), and at least the exhaust gas temperature (T e) 1 when providing One results, useful supercharger speed in estimating the state of the engine (n tc), scavenging pressure (P s), the scavenging temperature (T s), or the exhaust gas temperature (T e) You can get the diagnosis result about.
  • an abnormality control means for controlling the engine at the time of abnormality is provided based on the output of abnormality information, it is possible to prevent the abnormality of the engine from getting worse or leading to a failure by controlling the engine at the time of abnormality. ..
  • the human interface means when used as the information providing means to provide the abnormality information, the human interface means provides the abnormality information of the engine, so that the crew members and the like can respond promptly and appropriately in the event of an abnormality. It will be possible.
  • the transmission means for transmitting the abnormality information provided by the information providing means to another place is provided, the diagnosis result including the abnormality of the engine can be known in real time even at a place away from the engine.
  • the diagnosis result including the abnormality of the engine can be transmitted and provided online in real time.
  • Block diagram of engine abnormality diagnosis system Flowchart of abnormality diagnosis program of the same engine Schematic diagram of abnormality diagnosis of the engine Conceptual diagram of the same factor analysis
  • Diagram showing an example of a mathematical model of the engine Diagram showing the parameters of the engine model
  • a diagram showing the concept and calculation formula of the prediction (estimation) step and update (correction) step of the Kalman filter.
  • the engine abnormality diagnosis method, the engine abnormality diagnosis program, and the engine abnormality diagnosis system according to the embodiment of the present invention will be described below.
  • FIG. 1 is a block diagram of an engine abnormality diagnosis system.
  • the engine abnormality diagnosis system diagnoses an abnormality of the engine 1 mounted on a ship or the like by using a mathematical engine model 10.
  • the engine abnormality diagnosis system executes a condition input means 2 for inputting an initial state amount of the engine model 10, a state amount measuring means 3 for measuring the state of the engine 1 and obtaining a measured state amount, an engine abnormality diagnosis method or a program.
  • Computer 4 information providing means (human interface means: HMI) 5 that provides abnormality information including the diagnosis result of abnormality of engine 1 output from computer 4, abnormality that controls engine 1 at the time of abnormality based on output of abnormality information
  • the time control means 6, the transmission means 7 for transmitting the abnormality information provided by the information providing means 5 to other places, and the connection means 8 for connecting each device online are provided.
  • the engine model 10 is incorporated in the computer 4 from the beginning corresponding to the engine 1, but when a part or all of the engine 1 is changed to a different state in connection with maintenance of a ship, for example. Can be changed on the way via the condition input means 2 or the like.
  • connection means 8 is, for example, a router, a LAN, or the like, and connects the state quantity measuring means 3, the computer 4, and the information providing means 5 online. Thereby, the diagnosis result including the abnormality of the engine 1 can be transmitted and provided online in real time.
  • the online connection can be either wireless or wired via the connection means 8.
  • the transmission means 7 is provided in the computer 4. Further, the computer 4 includes a control unit 11, an initial state quantity acquisition unit 12, an engine state estimation unit 13, a measurement status quantity acquisition unit 14, a Kalman filtering unit 15, a repeat unit 16, a correlation calculation unit 17, and a factor analysis unit 18. It includes an abnormality detection unit 19, a machine learning application unit 20, an abnormality diagnosis unit 21, an engine model update unit 22, a main memory 23, an auxiliary memory 24, an engine model utilization unit 27, an output unit 28, and the like.
  • the auxiliary memory 24 is, for example, a hard disk or the like.
  • the engine model 10, the factor load space 25, and the engine failure database 26 are stored in the auxiliary memory 24.
  • the engine model 10 is pre-built based on the specifications and characteristics of the engine 1.
  • the factor load amount data is accumulated after detecting the abnormality of the engine 1. Further, in the engine failure database 26, the data collected by simulating the failure of the engine 1 by the simulation program is accumulated. If there is an engine state value at the time of actual failure, the engine state value can be accumulated in the engine failure database 26.
  • the condition input means 2 is a mouse, a keyboard, a touch panel, or the like.
  • the operator of the computer 4 or the user 101 inputs the initial state quantity using the condition input means 2.
  • the initial state quantity it is preferable to enter the fuel supply amount including the load of the engine 1 and the (Q p) fuel pump rack position (h p).
  • the input initial state quantity is transmitted to and acquired by the initial state quantity acquisition unit 12 of the computer 4. Further, when the operating conditions change to the extent that the engine model 10 changes, the condition input means 2 can input the condition signal.
  • the state quantity measuring means 3 is various sensors or the like. In the state quantity measuring means 3, it is preferable to obtain the rotation speed (ne ) of the engine 1 as the measured state quantity. As a result, the rotation speed ( ne ) of the engine 1, which is often measured as the engine 1, can be used for the abnormality diagnosis of the engine 1, and the measurement accuracy can be improved.
  • the measurement state quantity measured by the state quantity measuring means 3 In addition to this example of the engine 1 scavenging air pressure (P s), the exhaust gas temperature (T e), and the load of the engine 1 (Q p) and the like be.
  • the measured state quantity measured by the state quantity measuring means 3 is transmitted to the measurement status quantity acquisition unit 14 of the computer 4. Further, the measured state quantity may be processed based on the measured value, such as the average value of the measured value for a certain period or the difference from the average value, in addition to the measured value itself.
  • the output unit 28 outputs the diagnosis result of the engine 1 by the computer 4 to the transmission means 7, the information providing means (human interface means) 5, and the abnormality control means 6.
  • the information providing means 5, the abnormality control means 6, and the transmitting means 7 can be provided outside the computer 4.
  • the information providing means 5 can provide not only the diagnosis result of the abnormality of the engine 1 but also the information related to all the abnormalities including the abnormality detection information.
  • the transmission means 7 transmits the output diagnosis result to a device installed at a place different from the computer 4 via the connection means 8 via a wired or wireless LAN. As a result, it is possible to know the abnormality information including the diagnosis result of the abnormality of the engine 1 in real time even at a place away from the engine 1. Note that FIG.
  • the information providing means 5 transmits the abnormality information to the bridge 100 through the inboard LAN.
  • the information providing means (human interface means) 5 is, for example, a monitor, a speaker, or the like.
  • the information providing means 5 includes a storage means for temporarily storing abnormal information, a transfer means for transferring the abnormal information to a smartphone, and other means related to the provision of all kinds of information.
  • the abnormality control means 6 controls the engine 1 automatically or by an operation from the crew at the time of abnormality based on the abnormality detection result or the abnormality diagnosis result. By controlling the engine 1 at the time of abnormality, it is possible to prevent the abnormality of the engine 1 from getting worse or leading to a failure.
  • the information or signal provided to the abnormal time control means 6 may be raw information or signal of the information provided to the information providing means 5.
  • FIG. 2 is a flowchart of an engine abnormality diagnosis method
  • FIG. 3 is a schematic diagram of engine abnormality diagnosis. Since the engine abnormality diagnosis method can be provided as a program, it will be described below that the engine abnormality diagnosis program causes a computer to execute each step in the engine abnormality diagnosis method.
  • the control unit 11 reads the engine model 10 stored in the auxiliary memory 24 (engine model reading step S1). After the engine model reading step S1, the initial state quantity acquisition unit 12 acquires the initial state quantity of the engine model 10 input by the condition input means 2 (initial state quantity acquisition step S2).
  • the control unit 11 applies the initial state quantity to the engine model 10 read by the engine model utilization unit 27 and utilizes the engine model 10 (engine model utilization step S3).
  • the engine model 10 is initially set and updated. It is necessary to apply an arbitrary initial state amount in order to use the engine model 10, but in the engine model utilization step S3 that utilizes the engine model 10, the initial state amount (for example, load or fuel amount) that is as close to the current state as possible is applied. If applied, the engine model 10 can be utilized such that the calculation time in the engine model 10 is reduced and the accuracy of the estimation calculation by the engine model 10 is improved.
  • the engine model 10 is updated by inputting the residual in the Kalman filtering step S6 or utilizing the result of processing the acquired measurement state quantity in the engine model utilization step S3, and the engine model 10 is the actual engine. It will be faithful to 1 and will be utilized.
  • the engine state estimation unit 13 calculates the state of the engine 1 based on the initial state amount in the engine model 10 and obtains the estimated state amount (engine state estimation step S4). Further, the measurement state quantity acquisition unit 14 acquires the measurement status quantity of the engine 1 obtained by the measurement by the status quantity measuring means 3 (measurement status quantity acquisition step S5).
  • the Kalman filtering unit 15 inputs the residual between the acquired measured state quantity and the calculated estimated state quantity to the non-linear Kalman filter (Kalman filtering step S6).
  • Kalman filtering step S6 state estimation and model parameter estimation are performed.
  • the engine model update unit 22 determines whether or not the engine model 10 needs to be updated (model update determination step S7).
  • model update determination step S7 determines "Yes" in the model update determination step S7, that is, it is necessary to update the engine model 10
  • the result obtained in the Kalman filtering step S6 or the acquired measurement state quantity is used.
  • the processed result is applied to the engine model utilization step S3, and the engine model 10 is updated by updating the model parameters (model update step S8).
  • the determination in the model update determination step S7 is performed by setting a threshold value in the model parameter in advance.
  • the repeat unit 16 determines whether or not a predetermined time has elapsed. (Time lapse determination step S9).
  • the repeating unit 16 repeats k, k + 1, k + 2, ..., Divided by a time such as 0.1 seconds.
  • the repeat unit 16 determines "No" in the time lapse determination step S9, that is, when it is determined that a predetermined time has not elapsed, the repeat unit 16 applies the Kalman gain obtained by inputting to the nonlinear Kalman filter to the engine model 10.
  • the engine state estimation step S4, the measurement state quantity acquisition step S5, and the Kalman filtering step S6 are repeated (repeated step S10).
  • the two functions of the Kalman filter, state estimation and model parameter estimation are used to repeat the update and state estimation of the engine model 10.
  • the tracking filter can be used to identify the model parameters, and the Kalman filter can only estimate the state.
  • the nonlinear Kalman filter is preferably an unsented Kalman filter or an extended Kalman filter. As a result, the Kalman gain can be made more appropriate for the engine 1 which is a nonlinear system, and the calculation accuracy of the estimated state quantity can be improved.
  • the correlation calculation unit 17 calculates the correlation of the residual obtained at the time of input to the nonlinear Kalman filter. (Correlation calculation step S11).
  • the calculation of the residual correlation in the correlation calculation step S11 is preferably performed based on the correlation matrix. This makes it possible to improve the calculation accuracy of the residual correlation by a simple calculation. It is also possible to calculate the correlation by using the acquired measurement state quantity instead of the residual obtained at the time of input to the nonlinear Kalman filter.
  • the factor analysis unit 18 performs factor analysis on the correlation of the residuals to obtain the factor loading (factor analysis step S12). Instead of the correlation of the residual, the factor analysis can be performed using the correlation of the acquired measurement state quantity to obtain the factor loading amount.
  • the abnormality detection unit 19 calculates the factor score from the factor loading amount and detects the abnormality (abnormality detection step S13). The abnormality detected in the abnormality detection step S13 can also be output as abnormality information.
  • it is preferable to derive the factor loading by performing singular value decomposition (SVD) based on the covariance matrix as the correlation matrix. As a result, it is possible to extract what is essentially important regardless of the shape of the covariance matrix and detect the abnormality of the engine 1 at an early stage.
  • singular value decomposition SVD
  • the factor loading amount obtained in the factor analysis step S12 is accumulated in the factor loading amount space 25 (factor loading amount accumulation step S14).
  • the machine learning application unit 20 reads the data stored in the engine failure database 26 (engine failure data read step S15) and applies the factor load to machine learning (machine learning application step S16).
  • Machine learning uses a self-organizing map (SOM). Thereby, the cause of the abnormality of the engine 1 can be clearly classified by using SOM which is unsupervised machine learning.
  • SVM Small vector machine
  • Fuzzy C-means or the like can be used in addition to SOM.
  • the abnormality diagnosis unit 21 diagnoses an abnormality based on machine learning (abnormality diagnosis step S17).
  • the output unit 28 After the abnormality detection step S13 and the abnormality diagnosis step S17, the output unit 28 outputs the abnormality information including the abnormality diagnosis result of the engine 1 (output step S18). In the output of the abnormality information in the output step S18, the abnormality detection information and the accompanying information can be included in addition to the abnormality diagnosis result.
  • the output destinations are the information providing means (human interface means) 5, the abnormality control means 6, and the transmitting means 7.
  • FIG. 4 is a conceptual diagram of factor analysis.
  • Factor analysis is common to all the measurement values (measurement state amount) y m, is to look for the relationship is hidden expressed by a im (can not be measured) variable factor F. This is expressed by the following equation (1) using the linear relation parameter a.
  • f is common factor
  • a im is linear coefficients
  • remaining u i is the measurement error or noise factors that can not be explained.
  • Individual measurements y m is linked in a linear common factor f of a number.
  • the linear coefficient a im is called a factor loading.
  • FIG. 5 is a diagram showing an example of factor analysis using measurement data.
  • the measurement examples of the main factors and factor loadings A according to a specific track changes in the measured value the measured value y 1, y 2, across y m for example scavenging air pressure of the engine 1, exhaust gas temperature, represents across engine load, create a matrix collectively a section of a certain time (e.g., 0 ⁇ t period) (Fig. 5 ( a)).
  • the covariance matrix of this matrix is calculated and divided by the standard deviation of each variable to obtain R (FIG. 5 (b)).
  • R is decomposed into singular values (SVD) to obtain the singular value S.
  • the first factor loading amount A is obtained (FIG. 5 (c)).
  • the factor score D (index) is obtained by adding the squares of each first factor loading amount a and dividing by the sum of the variances (FIG. 5 (d)).
  • FIG. 6A and 6B are diagrams showing an example of factor scores
  • FIG. 6A shows a raw factor score F1 (D1)
  • FIG. 6B shows a filtered factor score F1 (D1).
  • FIG. 7 is a diagram showing changes in the factor load of each measured value
  • FIG. 7 (a) is a scavenging pressure
  • FIG. 7 (b) is a turbocharger rotation speed
  • FIG. 7 (c) is an exhaust gas temperature.
  • FIG. 7 (d) is a scavenging temperature.
  • the information contained in the factor loading matrix A represents an abnormality due to some change in the propulsion system.
  • the factor loading (row of the matrix A) represents the strength of the relationship between the propulsion system parameters as an element of error, and represents the characteristics of the cause of the abnormality. Therefore, the factor load is machine-learned using a machine learning algorithm, for example, a self-organizing map (SOM), etc., and is used for classifying the cause of an accident in the propulsion system.
  • the factor score D1 is used for early detection of abnormalities.
  • a Kalman filter observer is used as the Kalman filtering unit 15, and here, factor analysis is performed using the residual between the measured state quantity and the estimated state quantity. That is, in the present invention, as another method, instead of the measured value (measured state amount) Y, the residual E uniquely calculated by the Kalman filtering unit 15 between the measured state amount and the estimated state amount by the engine state estimation unit 13 is used. Perform the same factor analysis as above.
  • This residual E means a deviation from the estimated state quantity, and when the estimated state quantity is considered to be a normal state, it reflects the deviation from the normal state. Therefore, it is possible to detect that some abnormality has occurred in the engine 1 based on the residual E.
  • the Kalman filter observer is based on the digital twin engine model 10.
  • the residual E between the measured state quantity acquired by the dynamic process of the engine 1 and the estimated state quantity calculated by the process of the mathematical engine model 10 based on the initial state quantity is input to the nonlinear Kalman filter. ..
  • Kalman gain is obtained.
  • the Kalman gain is applied to the engine model 10 and is used for the process control of the mathematical engine model 10. In this way, by monitoring the engine state using the engine model 10 as the digital twin model of the engine 1, it is possible to detect the failure of the engine 1 at an early stage and diagnose the cause.
  • FIG. 8 is a diagram showing an example of a mathematical model of an engine.
  • the actual fuel supply system in the engine 1 fuel pump rack position (h p)
  • a relational diagram of the measurement system such as the engine revolution speed (n e)
  • the measurement points and measurement values are shown.
  • the right side in FIG. 8 shows an engine model 10 which is a mathematical model of the engine 1.
  • the state quantity measuring means 3 is shown by the enclosing English characters. "T” surrounded by ⁇ is a thermometer, "P” surrounded by ⁇ is a pressure gauge, "n” surrounded by ⁇ is a tachometer, and "Q" surrounded by ⁇ is a shaft horsepower meter.
  • Load fluctuations are measured with an axial horsepower meter.
  • the measured value is measured at each step (k, k + 1, k + 2, ...), And is calculated (estimated) at each step by the engine model 10.
  • the rotation speed ( ne ) of the engine 1 that can be measured with the highest accuracy is acquired, the Kalman gain is calculated, and the engine model 10 is modified. This is Kalman filtering.
  • the behavior in the normal state can be represented by a non-linear state-space model.
  • the non-linear state space model is composed of an equation represented by the equation of state X, each parameter represented by the state quantity x, and an input represented by the input quantity u.
  • FIG. 9 is a diagram showing parameters of the engine model.
  • the model parameters, the load of the engine such as a propeller torque (Q p), engine torque (Q e), the moment of inertia (I e, I tc), the fuel pump rack position (h p), the rotational speed of the engine (n e ), Supercharger rotation speed ( ntc ), atmospheric pressure (P a ), atmospheric temperature (T a ), scavenging pressure (P s ), scavenging temperature (T s ), maximum compression pressure in the cylinder (P c ), Maximum combustion pressure in cylinder (P z ), average effective pressure in cylinder (P i ), scavenging receiver volume (V a r ), exhaust receiver volume (V e r ), thermodynamic constants (R a , R e , k e, C pe, C pa ), the cooling water temperature (T w), the compressor outlet temperature (T c), the exhaust gas pressure (P e),
  • FIG. 10 is a diagram showing the concept and calculation formula of the Kalman filter prediction (estimation) step (Kalman filtering step S6) and the update (correction) step (model update step S8).
  • Kalman filtering step S6 the update (correction) step
  • model update step S8 the update (correction) step.
  • an error occurs between the estimated state quantity by the engine model 10 and the measured state quantity.
  • the unsented Kalman filter which is a nonlinear Kalman filter, calculates this error and corrects it with the measured state quantity to bring the estimated state quantity as close to the correct value as possible.
  • FIG. 10 shows the basics of the unsented Kalman filter.
  • the behavior of the propulsion system is repeatedly expressed by the calculated error at each discrete sampling time k, taking into account the uncertainty of measurement and modeling (Equation 2 below).
  • the error distribution is considered to be a 0 average normal distribution.
  • the error generated by the Kalman filter is used for factor analysis.
  • FIG. 11 is a diagram showing the relationship between the Kalman filter and factor analysis.
  • the Kalman filter observer is based on the engine model 10 of the digital twin, the covariance estimation is performed every time, and the Kalman gain is used as an indicator of which state quantity should be corrected.
  • FIG. 12 is a diagram showing an example of abnormality detection based on a factor score.
  • FIG. 12 shows an example in which a blockage simulation experiment of the supercharger suction filter of the actual engine 1 was performed and an abnormality was quickly detected by the engine abnormality diagnosis system.
  • the horizontal axis of the graph is the number of samples, and one time is 0.1 seconds.
  • the turbocharger suction filter is gradually blocked and the pressure loss increases, but the factor score F1 rises sharply at the initial stage when the pressure loss starts to increase, so that an abnormality can be detected.
  • the computer 4 can be used to provide abnormality information including the result of early detection of the abnormality of the engine 1 and diagnosis of the cause thereof.
  • Each component and peripheral means of the computer 4 can be externally attached or incorporated as appropriate, and the computer 4 may be divided into roles among a plurality of computers, or a part of the computer 4 may be a discrete circuit. It is possible.
  • the present invention contributes to safe and efficient operation because, for example, early detection and diagnosis of an abnormality in the engine of a vessel in service can be performed, and the diagnosis result including the abnormality of the engine can be known in real time even at a remote location. It can also be used for early detection and diagnosis of abnormalities in engines other than ships.
  • Engine 2 Condition input means 3 State quantity measuring means 4 Computer 5 Information providing means (human interface means) 6 Abnormal time control means 7 Transmission means 8 Connection means 10 Engine model S2 Initial state quantity acquisition step S3 Engine model utilization step S4 Engine status estimation step S5 Measurement status quantity acquisition step S6 Kalman filtering step S8 Model update step S10 Repeat step S11 Correlation Calculation step S12 Factor analysis step S13 Abnormality detection step S16 Machine learning application step S17 Abnormality diagnosis step S18 Output step

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Mechanical Engineering (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Physics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Combined Controls Of Internal Combustion Engines (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

This engine abnormality diagnosis method, engine abnormality diagnosis program, and engine abnormality diagnosis system, with which early detection and cause diagnosis of an engine abnormality are performed, involve execution of a step S2 for acquiring an initial state quantity of an engine model 10, a step S3 for applying the initial state quantity and utilizing the engine model 10, a step S4 for obtaining an estimated state quantity using the engine model 10, a step S5 for acquiring a measured state quantity of an engine 1, a step S6 for subjecting the residual error between the measured state quantity and the estimated state quantity to a nonlinear Kalman filter, a step S10 for applying a Kalman gain to the engine model 10 and repeating steps S4–S6, a step S11 for calculating a measured state quantity or a residual error correlation, a step S12 for performing a factor analysis on the measured state quantity or the residual error correlation to derive a factor load quantity, a step S13 for calculating a factor score from the factor load quantity and detecting an abnormality, a step S16 for applying the factor load quantity to machine learning, a step S17 for diagnosing the abnormality on the basis of the machine learning, and a step S18 for outputting the result of diagnosis.

Description

エンジンの異常診断方法、エンジンの異常診断プログラム、及びエンジンの異常診断システムEngine abnormality diagnosis method, engine abnormality diagnosis program, and engine abnormality diagnosis system
　本発明は、エンジンの異常を数学的なエンジンモデルを用いて異常診断する、エンジンの異常診断方法、エンジンの異常診断プログラム、及びエンジンの異常診断システムに関する。 The present invention relates to an engine abnormality diagnosis method, an engine abnormality diagnosis program, and an engine abnormality diagnosis system for diagnosing an engine abnormality using a mathematical engine model.
　船舶等に搭載されているエンジンの異常を早期に検知することは安全面や効率面等の観点から重要である。
　ここで、特許文献１には、エンジンのある作動点で最適に設計されたカルマンフィルタ・ゲインと、実機エンジンを忠実にモデル化した非線形ダイナミックシミュレーション・モデルとから成る一定ゲイン拡張カルマンフィルタ（ＣＧＥＫＦ）を利用して、観測変数と観測変数推定値との間に差が生じる場合はモデルの状態変数の一部分を成すチューニングパラメータをその差が最小となるように調整することにより、その非線形ダイナミックシミュレーション・モデルを常に実機に忠実なモデルとするガスタービンエンジンの性能推定システムが開示されている。また、段落００１０には、エンジン使用による経年変化や損傷による性能劣化を、エンジンの非線形ダイナミックシミュレーション・モデルと推定フィルタにより常に推定・検出・監視するエンジン性能推定システムおよび方法であることが記載されている。
　また、特許文献２には、航空機の飛行中に航空機の飛行パラメータをリアルタイムで決定するためのシステムであって、航空機の推定されるべき少なくとも２個の事前選択された飛行パラメータ間の相互依存関係を定める飛行力学方程式に基づき構成されて、航空機の飛行中に、選択された飛行パラメータの合同推定値を供するように形成された拡張カルマンフィルタを含むシステムが開示されている。
　また、特許文献３には、プロセスを制御するのに用いられる制御信号をコントローラが生成できるようにするためにコントローラの各実行サイクル中にコントローラに新しいプロセス変数推定を提供しながら、遅い又は間欠的なプロセスフィードバック信号からプロセス変数値の推定を生じるように構成されるカルマンフィルタが開示されている。 It is important from the viewpoint of safety and efficiency to detect an abnormality of the engine mounted on a ship or the like at an early stage.
Here, Patent Document 1 uses a constant gain extended Kalman filter (CGEKF) consisting of a Kalman filter gain optimally designed at a certain operating point of the engine and a nonlinear dynamic simulation model that faithfully models the actual engine. Then, if there is a difference between the observed variable and the observed variable estimate, the nonlinear dynamic simulation model can be adjusted by adjusting the tuning parameters that form part of the state variable of the model so that the difference is minimized. A performance estimation system for a gas turbine engine, which is always a model faithful to the actual machine, is disclosed. Further, paragraph 0010 describes an engine performance estimation system and method that constantly estimates, detects, and monitors performance deterioration due to aging and damage due to engine use by using a nonlinear dynamic simulation model of the engine and an estimation filter. There is.
Further, Patent Document 2 is a system for determining the flight parameters of an aircraft in real time during the flight of the aircraft, and the interdependence between at least two preselected flight parameters of the aircraft to be estimated. Disclosed is a system comprising an extended Kalman filter configured on the basis of a flight dynamics equation to provide a congruent estimate of selected flight parameters during flight of an aircraft.
Also in Patent Document 3, slow or intermittent while providing the controller with new process variable estimates during each execution cycle of the controller to allow the controller to generate control signals used to control the process. A Kalman filter configured to generate an estimation of a process variable value from a different process feedback signal is disclosed.
特開２００９－６８３５９号公報Japanese Unexamined Patent Publication No. 2009-68359 特開２０１３－４９４０８号公報Japanese Unexamined Patent Publication No. 2013-49408 特開２０１８－８８２８９号公報Japanese Unexamined Patent Publication No. 2018-88289
　特許文献１は、段落０００５に、「計測可能項目と計測不能項目との間の相関関係を変化させる経年劣化や損傷等の影響が十分に反映された高精度な非線形ダイナミックシミュレーション・モデルとすることが可能となる。」と記載されているように、非線形ダイナミックシミュレーション・モデルを逐次チューニングし、エンジンの計測不能項目を計測可能項目から推定し、その推定結果を用いてガスタービンエンジンを制御するものであり、エンジンの異常の早期検知とその原因の診断を行うものではない。　
　特許文献２は、飛行力学方程式によって相互に関連付けられた少なくとも２個の飛行パラメータを拡張カルマンフィルタによって同時に推定して航空機の飛行制御に用いるものであり、エンジンの異常の早期検知とその原因の診断を行うものではない。
　特許文献３は、化学プロセスや石油プロセス等のプロセス制御システムにおけるカルマンフィルタの改善に関するものであり、エンジンの異常の早期検知とその原因の診断を行うものではない。
　そこで本発明は、エンジンの異常の早期検知とその原因の診断を行うことができるエンジンの異常診断方法、エンジンの異常診断プログラム、及びエンジンの異常診断システムを提供することを目的とする。 In paragraph 0005, Patent Document 1 provides a high-precision nonlinear dynamic simulation model that sufficiently reflects the effects of aging deterioration, damage, etc. that change the correlation between measurable items and non-measurable items. The non-linear dynamic simulation model is sequentially tuned, the unmeasurable items of the engine are estimated from the measurable items, and the gas turbine engine is controlled using the estimation results. Therefore, it does not detect an abnormality in the engine at an early stage and diagnose its cause.
Patent Document 2 simultaneously estimates at least two flight parameters associated with each other by flight dynamics equations by an extended Kalman filter and uses them for flight control of an aircraft, and early detection of engine abnormality and diagnosis of the cause thereof. It's not something to do.
Patent Document 3 relates to an improvement of a Kalman filter in a process control system such as a chemical process or a petroleum process, and does not perform early detection of an abnormality in an engine and diagnosis of the cause thereof.
Therefore, an object of the present invention is to provide an engine abnormality diagnosis method, an engine abnormality diagnosis program, and an engine abnormality diagnosis system capable of early detection of an engine abnormality and diagnosis of the cause thereof.
　請求項１記載に対応したエンジンの異常診断方法においては、エンジンの異常を、数学的なエンジンモデルを用いて異常診断するプログラムであって、エンジンモデルの初期状態量を取得する初期状態量取得ステップと、エンジンモデルに初期状態量を適用しエンジンモデルを活用するエンジンモデル活用ステップと、エンジンモデルで初期状態量に基づいてエンジンの状態を計算し推定状態量を得るエンジン状態推定ステップと、エンジンの計測状態量を取得する計測状態量取得ステップと、取得した計測状態量と計算した推定状態量との残差を非線形カルマンフィルタにかけるカルマンフィルタリングステップと、非線形カルマンフィルタにかけて得られたカルマンゲインをエンジンモデルに適用し、エンジン状態推定ステップと、計測状態量取得ステップと、カルマンフィルタリングステップを繰り返す繰返ステップと、非線形カルマンフィルタにかけたときの計測状態量、又は残差の相関を計算する相関計算ステップと、計測状態量、又は残差の相関に対して因子分析し因子負荷量を求める因子分析ステップと、因子負荷量から因子スコアを計算し異常を検知する異常検知ステップと、因子負荷量を機械学習に適用する機械学習適用ステップと、機械学習に基づいて異常を診断する異常診断ステップと、エンジンの異常の診断結果を含む異常情報を出力する出力ステップとを実行することを特徴とする。
　請求項１に記載の本発明によれば、計測状態量、又は計測状態量と計算した推定状態量との残差を利用して因子分析にかけることにより、計算した因子スコアに基づきエンジンの異常を早期に検知することができる。また、因子負荷量を機械学習に適用しエンジンの異常の原因を診断することができる。 The engine abnormality diagnosis method according to claim 1 is a program for diagnosing an engine abnormality using a mathematical engine model, and is an initial state quantity acquisition step for acquiring the initial state quantity of the engine model. And the engine model utilization step that applies the initial state quantity to the engine model and utilizes the engine model, the engine state estimation step that calculates the engine state based on the initial state quantity in the engine model and obtains the estimated state quantity, and the engine The Kalman filtering step for acquiring the measured state quantity, the Kalman filtering step for applying the residual between the acquired measured state quantity and the calculated estimated state quantity to the nonlinear Kalman filter, and the Kalman gain obtained by applying the nonlinear Kalman filter to the engine model. Apply, engine state estimation step, measurement state quantity acquisition step, repeat step that repeats Kalman filtering step, measurement state quantity when applied to nonlinear Kalman filter, correlation calculation step to calculate the correlation of residual, and measurement. A factor analysis step that performs factor analysis on the correlation of state functions or residuals to obtain factor loadings, an abnormality detection step that calculates factor scores from factor loadings and detects abnormalities, and factor loadings are applied to machine learning. It is characterized by executing a machine learning application step, an abnormality diagnosis step for diagnosing an abnormality based on machine learning, and an output step for outputting abnormality information including an engine abnormality diagnosis result.
According to the first aspect of the present invention, an abnormality of the engine is performed based on the calculated factor score by performing factor analysis using the measured state quantity or the residual between the measured state quantity and the calculated estimated state quantity. Can be detected at an early stage. In addition, the factor load can be applied to machine learning to diagnose the cause of engine abnormality.
　請求項２記載の本発明は、カルマンフィルタリングステップで得られた結果、又は取得した計測状態量を処理した結果をエンジンモデル活用ステップに適用して、エンジンモデルを更新するモデル更新ステップをさらに実行することを特徴とする。
　請求項２に記載の本発明によれば、エンジンモデルを更新して推定状態量の計算精度を常に高い状態に保つことができる。 The present invention according to claim 2 further executes a model update step for updating an engine model by applying the result obtained in the Kalman filtering step or the result of processing the acquired measurement state quantity to the engine model utilization step. It is characterized by that.
According to the second aspect of the present invention, the engine model can be updated to keep the calculation accuracy of the estimated state quantity always high.
　請求項３記載の本発明は、相関計算ステップにおける計測状態量、又は残差の相関の計算は、相関行列に基づいて行うことを特徴とする。
　請求項３に記載の本発明によれば、簡単な計算により計測状態量、又は残差の相関の計算精度を向上させることができる。 The present invention according to claim 3 is characterized in that the calculation of the correlation of the measured state quantity or the residual in the correlation calculation step is performed based on the correlation matrix.
According to the third aspect of the present invention, the calculation accuracy of the measured state quantity or the correlation of the residual can be improved by a simple calculation.
　請求項４記載の本発明は、因子分析ステップにおいて、相関行列としての共分散行列に基づいて特異値分解(ＳＶＤ)をして因子負荷量を導出することを特徴とする。
　請求項４に記載の本発明によれば、共分散行列の形に捉われず本質的に重要なものを抽出し、エンジンの異常を早期に検知することができる。 The present invention according to claim 4 is characterized in that, in the factor analysis step, a factor loading is derived by performing singular value decomposition (SVD) based on a covariance matrix as a correlation matrix.
According to the fourth aspect of the present invention, it is possible to extract an essentially important one regardless of the shape of the covariance matrix and detect an abnormality of the engine at an early stage.
　請求項５記載の本発明は、機械学習は、自己組織化マップ(ＳＯＭ)を用いることを特徴とする。
　請求項５に記載の本発明によれば、教師なし機械学習である自己組織化マップ（ＳＯＭ）を利用してエンジンの異常の原因を分類することができる。 The present invention according to claim 5 is characterized in that machine learning uses a self-organizing map (SOM).
According to the fifth aspect of the present invention, the cause of an engine abnormality can be classified by using a self-organizing map (SOM) which is unsupervised machine learning.
　請求項６記載の本発明は、初期状態量取得ステップで取得する初期状態量は、エンジンの負荷(Ｑｐ)と燃料ポンプラック位置(ｈｐ)を含む燃料供給量であることを特徴とする。
　請求項６に記載の本発明によれば、エンジンの状態を推定する上で重要であるエンジンの負荷(Ｑｐ)と燃料ポンプラック位置(ｈｐ)を含む燃料供給量に基づいて、推定状態量を得ることができる。 According to a sixth aspect of the invention, the initial state quantity acquired in the initial state quantity acquisition step, characterized in that it is a fuel supply amount including load (Q p) fuel pump rack position (h p) of the engine ..
According to the present invention described in claim 6, based on the fuel supply amount including the important engine in order to estimate the state of the engine load (Q p) fuel pump rack position (h p), the estimated state You can get the quantity.
　請求項７記載の本発明は、計測状態量取得ステップで取得する計測状態量は、エンジンの回転数(ｎｅ)であることを特徴とする。
　請求項７に記載の本発明によれば、エンジンとして計測する機会の多いエンジンの回転数(ｎｅ)を、エンジンの異常診断に用いることができ、計測精度も高くすることができる。 The present invention according to claim 7 is characterized in that the measured state quantity acquired in the measured state quantity acquisition step is the engine speed ( ne ).
According to the seventh aspect of the present invention, the engine speed ( ne ), which is often measured as an engine, can be used for engine abnormality diagnosis, and the measurement accuracy can be improved.
　請求項８記載の本発明は、エンジン状態推定ステップの推定状態量として、過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、及び排気ガス温度(Ｔｅ)を得ることを特徴とする。
　請求項８に記載の本発明によれば、エンジンの状態を推定する上で有用な過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、又は排気ガス温度(Ｔｅ)についての診断結果を得ることができる。 The present invention of claim 8, wherein, as the estimated state quantity of the engine state estimating step, the supercharger rotational speed (n tc), scavenging pressure (P s), the scavenging temperature (T s), and the exhaust gas temperature (T e ) Is obtained.
According to the present invention described in claim 8, useful supercharger speed in estimating the state of the engine (n tc), scavenging pressure (P s), the scavenging temperature (T s), or the exhaust gas temperature (T e) diagnostics for can be obtained.
　請求項９記載に対応したエンジンの異常診断プログラムにおいては、エンジンの異常を、数学的なエンジンモデルを用いて異常診断するプログラムであって、コンピュータに、エンジンの異常診断方法における初期状態量取得ステップ、エンジンモデル活用ステップ、エンジン状態推定ステップ、計測状態量取得ステップ、カルマンフィルタリングステップ、繰返ステップ、相関計算ステップ、因子分析ステップ、異常検知ステップ、機械学習適用ステップ、異常診断ステップ、及び出力ステップを実行させることを特徴とする。
　請求項９に記載の発明によれば、計測状態量、又は計測状態量と計算した推定状態量との残差を利用して因子分析にかけることにより、計算した因子スコアに基づきエンジンの異常を早期に検知することができる。また、因子負荷量を機械学習に適用しエンジンの異常の原因を診断することができる。 The engine abnormality diagnosis program corresponding to claim 9 is a program for diagnosing an engine abnormality using a mathematical engine model, and is a step of acquiring an initial state amount in an engine abnormality diagnosis method using a computer. , Engine model utilization step, engine state estimation step, measurement state quantity acquisition step, Kalman filtering step, repeat step, correlation calculation step, factor analysis step, abnormality detection step, machine learning application step, abnormality diagnosis step, and output step. It is characterized by being executed.
According to the invention of claim 9, an abnormality of the engine is detected based on the calculated factor score by performing factor analysis using the measured state quantity or the residual between the measured state quantity and the calculated estimated state quantity. It can be detected early. In addition, the factor load can be applied to machine learning to diagnose the cause of engine abnormality.
　請求項１０記載に対応したエンジンの異常診断システムにおいては、エンジンと、エンジンモデルの初期状態量を入力する条件入力手段と、エンジンの状態を計測し計測状態量を得る状態量計測手段と、エンジンの異常診断方法、又はエンジンの異常診断プログラムを実行するコンピュータと、コンピュータより出力されるエンジンの異常の診断結果を含む異常情報を提供する情報提供手段とを備えたことを特徴とする。
　請求項１０に記載の本発明によれば、コンピュータを利用して、エンジンの異常の早期検知とその原因の診断を行った結果を含む異常情報を提供することができる。 In the engine abnormality diagnosis system corresponding to the tenth aspect, the engine, the condition input means for inputting the initial state amount of the engine model, the state amount measuring means for measuring the state of the engine and obtaining the measured state amount, and the engine. The present invention is characterized by comprising a computer for executing the abnormality diagnosis method of the engine or an abnormality diagnosis program of the engine, and an information providing means for providing abnormality information including the diagnosis result of the abnormality of the engine output from the computer.
According to the tenth aspect of the present invention, it is possible to use a computer to provide abnormality information including a result of early detection of an abnormality of an engine and diagnosis of the cause thereof.
　請求項１１記載の本発明は、コンピュータで、エンジンモデルの更新を行うことを特徴とする。
　請求項１１に記載の本発明によれば、コンピュータで、エンジンモデルを更新することにより推定状態量の計算精度の向上が容易にできる。 The present invention according to claim 11 is characterized in that the engine model is updated by a computer.
According to the eleventh aspect of the present invention, it is possible to easily improve the calculation accuracy of the estimated state quantity by updating the engine model with a computer.
　請求項１２記載の本発明は、状態量計測手段で、計測状態量としてエンジンの回転数(ｎｅ)を得ることを特徴とする。
　請求項１２に記載の本発明によれば、エンジンとして計測する機会の多いエンジンの回転数(ｎｅ)を、エンジンの異常診断に用いることができ、計測精度も高くすることができる。 The present invention according to claim 12 is characterized in that a state quantity measuring means obtains an engine speed (ne) as a measured state quantity.
According to the twelfth aspect of the present invention, the engine speed ( ne ), which is often measured as an engine, can be used for engine abnormality diagnosis, and the measurement accuracy can be improved.
　請求項１３記載の本発明は、情報提供手段で、異常の診断結果として、エンジンの過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、及び排気ガス温度(Ｔｅ)の少なくとも１つの結果を提供することを特徴とする。
　請求項１３に記載の本発明によれば、エンジンの状態を推定する上で有用な過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、又は排気ガス温度(Ｔｅ)についての診断結果を得ることができる。 The present invention according to claim 13 is an information providing means, and as a result of diagnosing an abnormality, an engine supercharger rotation speed ( ntc ), a scavenging pressure (P s ), a scavenging temperature (T s ), and an exhaust gas temperature. and providing at least one result of the (T e).
According to the present invention of claim 13, useful supercharger speed in estimating the state of the engine (n tc), scavenging pressure (P s), the scavenging temperature (T s), or the exhaust gas temperature (T e) diagnostics for can be obtained.
　請求項１４記載の本発明は、異常情報の出力に基づいて、異常時にエンジンを制御する異常時制御手段を備えたことを特徴とする。
　請求項１４に記載の本発明によれば、異常時にエンジンを制御することにより、エンジンの異常が悪化したり、故障に至ることを防止できる。 The present invention according to claim 14 is characterized in that an abnormality control means for controlling an engine at the time of abnormality is provided based on the output of abnormality information.
According to the 14th aspect of the present invention, by controlling the engine at the time of abnormality, it is possible to prevent the abnormality of the engine from getting worse or leading to a failure.
　請求項１５記載の本発明は、情報提供手段として、ヒューマンインターフェース手段を用いて異常情報を提供することを特徴とする。
　請求項１５に記載の本発明によれば、ヒューマンインターフェース手段からエンジンの異常情報が提供されることで、乗組員等は異常時に迅速かつ適切に対応することが可能となる。 The present invention according to claim 15 is characterized in that anomalous information is provided by using a human interface means as an information providing means.
According to the fifteenth aspect of the present invention, by providing the abnormality information of the engine from the human interface means, the crew and the like can quickly and appropriately respond to the abnormality.
　請求項１６記載の本発明は、情報提供手段で提供される異常情報を、他の箇所に送信する送信手段を備えたことを特徴とする。
　請求項１６に記載の本発明によれば、エンジンから離れた場所においてもリアルタイムにエンジンの異常を含む診断結果を知ることができる。 The present invention according to claim 16 is characterized by comprising a transmission means for transmitting anomalous information provided by the information providing means to another place.
According to the sixteenth aspect of the present invention, it is possible to know the diagnosis result including the abnormality of the engine in real time even at a place away from the engine.
　請求項１７記載の本発明は、状態量計測手段と、コンピュータと、情報提供手段とをオンラインで接続する接続手段を備えたことを特徴とする。
　請求項１７に記載の本発明によれば、エンジンの異常を含む診断結果を、オンラインでリアルタイムに伝送して提供することができる。 The present invention according to claim 17 is characterized by comprising a state quantity measuring means, a connecting means for connecting a computer and an information providing means online.
According to the 17th aspect of the present invention, a diagnostic result including an abnormality of an engine can be transmitted and provided online in real time.
　本発明のエンジンの異常診断プログラムによれば、計測状態量、又は計測状態量と計算した推定状態量との残差を利用して因子分析にかけることにより、計算した因子スコアに基づきエンジンの異常を早期に検知することができる。また、因子負荷量を機械学習に適用しエンジンの異常の原因を診断することができる。 According to the engine abnormality diagnosis program of the present invention, an engine abnormality is performed based on the calculated factor score by performing factor analysis using the measured state quantity or the residual between the measured state quantity and the calculated estimated state quantity. Can be detected at an early stage. In addition, the factor load can be applied to machine learning to diagnose the cause of engine abnormality.
　また、カルマンフィルタリングステップで得られた結果、又は取得した計測状態量を処理した結果をエンジンモデル活用ステップに適用して、エンジンモデルを更新するモデル更新ステップをさらに実行する場合は、エンジンモデルを更新して推定状態量の計算精度を常に高い状態に保つことができる。 In addition, when the result obtained in the Kalman filtering step or the result obtained by processing the acquired measurement state quantity is applied to the engine model utilization step and the model update step for updating the engine model is further executed, the engine model is updated. Therefore, the calculation accuracy of the estimated state quantity can always be kept high.
　また、相関計算ステップにおける残差の相関の計算は、相関行列に基づいて行う場合は、簡単な計算により計測状態量、又は残差の相関の計算精度を向上させることができる。 Further, when the residual correlation calculation in the correlation calculation step is performed based on the correlation matrix, the measurement state quantity or the residual correlation calculation accuracy can be improved by a simple calculation.
　また、因子分析ステップにおいて、相関行列としての共分散行列に基づいて特異値分解(ＳＶＤ)をして因子負荷量を導出する場合は、共分散行列の形に捉われず本質的に重要なものを抽出し、エンジンの異常を早期に検知することができる。 In addition, in the factor analysis step, when the factor loading is derived by singular value decomposition (SVD) based on the covariance matrix as the correlation matrix, it is essentially important regardless of the shape of the covariance matrix. Can be extracted and an abnormality in the engine can be detected at an early stage.
　また、機械学習は、自己組織化マップ(ＳＯＭ)を用いる場合は、教師なし機械学習である自己組織化マップ（ＳＯＭ）を利用してエンジンの異常の原因を分類することができる。 In addition, when machine learning uses a self-organizing map (SOM), the cause of engine abnormality can be classified using the self-organizing map (SOM), which is unsupervised machine learning.
　また、初期状態量取得ステップで取得する初期状態量は、エンジンの負荷(Ｑｐ)と燃料ポンプラック位置(ｈｐ)を含む燃料供給量である場合は、エンジンの状態を推定する上で重要であるエンジンの負荷(Ｑｐ)と燃料ポンプラック位置(ｈｐ)を含む燃料供給量に基づいて推定状態量を得ることができる。 The initial state quantity acquired in the initial state quantity acquisition step, if a fuel supply amount including load (Q p) fuel pump rack position (h p) of the engine, important for estimating the state of the engine it is possible to obtain an estimated state quantity based on the fuel supply amount including load (Q p) fuel pump rack position (h p) of the engine is.
　また、計測状態量取得ステップで取得する計測状態量は、エンジンの回転数(ｎｅ)である場合は、エンジンとして計測する機会の多いエンジンの回転数(ｎｅ)をエンジンの異常診断に用いることができ、計測精度も高くすることができる。 Further, the measurement state quantity acquired by measuring the state quantity obtaining step, when a rotational speed of the engine (n e), using the rotation speed of the large engines opportunity to measure the engine (n e) in the abnormality diagnosis of the engine It is possible to improve the measurement accuracy.
　また、エンジン状態推定ステップの推定状態量として、過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、及び排気ガス温度(Ｔｅ)を得る場合は、エンジンの状態を推定する上で有用な過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、又は排気ガス温度(Ｔｅ)についての診断結果を得ることができる。 Further, as the estimated state quantity of the engine state estimating step, the supercharger rotational speed (n tc), scavenging pressure (P s), the scavenging temperature (T s), and the case of obtaining an exhaust gas temperature (T e), the engine state useful supercharger speed in estimating the (n tc), it is possible to obtain a diagnosis result for scavenging air pressure (P s), the scavenging temperature (T s), or the exhaust gas temperature (T e) ..
　また、本発明のエンジンの異常診断プログラムによれば、計測状態量、又は計測状態量と計算した推定状態量との残差を利用して因子分析にかけることにより、計算した因子スコアに基づきエンジンの異常を早期に検知することができる。また、因子負荷量を機械学習に適用しエンジンの異常の原因を診断することができる。 Further, according to the abnormality diagnosis program of the engine of the present invention, the engine is based on the calculated factor score by performing factor analysis using the measured state quantity or the residual between the measured state quantity and the calculated estimated state quantity. Abnormality can be detected at an early stage. In addition, the factor load can be applied to machine learning to diagnose the cause of engine abnormality.
　また、本発明のエンジンの異常診断システムによれば、コンピュータを利用して、エンジンの異常の早期検知とその原因の診断を行った結果を含む異常情報を提供することができる。 Further, according to the engine abnormality diagnosis system of the present invention, it is possible to provide abnormality information including the result of early detection of engine abnormality and diagnosis of the cause thereof by using a computer.
　また、コンピュータで、エンジンモデルの更新を行う場合は、コンピュータで、エンジンモデルを更新することにより推定状態量の計算精度の向上が容易にできる。 Also, when updating the engine model on a computer, it is possible to easily improve the calculation accuracy of the estimated state quantity by updating the engine model on the computer.
　また、状態量計測手段で、計測状態量としてエンジンの回転数(ｎｅ)を得る場合は、エンジンとして計測する機会の多いエンジンの回転数(ｎｅ)をエンジンの異常診断に用いることができ、計測精度も高くすることができる。 In a state quantity measuring means, the case of obtaining a rotational speed of the engine (n e) as the measured state variables can be used in many engines opportunity to measure the engine rotational speed (n e) in the abnormality diagnosis of the engine , The measurement accuracy can also be improved.
　また、情報提供手段で、異常の診断結果として、エンジンの過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、及び排気ガス温度(Ｔｅ)の少なくとも１つの結果を提供する場合は、エンジンの状態を推定する上で有用な過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、又は排気ガス温度(Ｔｅ)についての診断結果を得ることができる。 Further, the information providing unit, as a diagnosis result of the abnormality, the supercharger rotation speed of the engine (n tc), scavenging pressure (P s), the scavenging temperature (T s), and at least the exhaust gas temperature (T e) 1 when providing One results, useful supercharger speed in estimating the state of the engine (n tc), scavenging pressure (P s), the scavenging temperature (T s), or the exhaust gas temperature (T e) You can get the diagnosis result about.
　また、異常情報の出力に基づいて、異常時にエンジンを制御する異常時制御手段を備えた場合は、異常時にエンジンを制御することにより、エンジンの異常が悪化したり、故障に至ることを防止できる。 Further, when an abnormality control means for controlling the engine at the time of abnormality is provided based on the output of abnormality information, it is possible to prevent the abnormality of the engine from getting worse or leading to a failure by controlling the engine at the time of abnormality. ..
　また、情報提供手段として、ヒューマンインターフェース手段を用いて異常情報を提供する場合は、ヒューマンインターフェース手段からエンジンの異常情報が提供されることで、乗組員等は異常時に迅速かつ適切に対応することが可能となる。 In addition, when the human interface means is used as the information providing means to provide the abnormality information, the human interface means provides the abnormality information of the engine, so that the crew members and the like can respond promptly and appropriately in the event of an abnormality. It will be possible.
　また、情報提供手段で提供される異常情報を、他の箇所に送信する送信手段を備えた場合は、エンジンから離れた場所においてもリアルタイムにエンジンの異常を含む診断結果を知ることができる。 Further, when the transmission means for transmitting the abnormality information provided by the information providing means to another place is provided, the diagnosis result including the abnormality of the engine can be known in real time even at a place away from the engine.
　また、状態量計測手段と、コンピュータと、情報提供手段とをオンラインで接続する接続手段を備えた場合は、エンジンの異常を含む診断結果を、オンラインでリアルタイムに伝送して提供することができる。 Further, when the state quantity measuring means, the computer, and the connecting means for connecting the information providing means online are provided, the diagnosis result including the abnormality of the engine can be transmitted and provided online in real time.
本発明の実施形態によるエンジンの異常診断システムのブロック図Block diagram of engine abnormality diagnosis system according to the embodiment of the present invention 同エンジンの異常診断プログラムのフローチャートFlowchart of abnormality diagnosis program of the same engine 同エンジンの異常診断の概要図Schematic diagram of abnormality diagnosis of the engine 同因子分析の概念図Conceptual diagram of the same factor analysis 同計測データを用いた因子分析の例を示す図The figure which shows the example of the factor analysis using the same measurement data 同因子スコアの例を示す図Diagram showing an example of the same factor score 同各計測値の因子負荷量の変化を示す図The figure which shows the change of the factor loading amount of the same measured value 同エンジンの数学モデルの例を示す図Diagram showing an example of a mathematical model of the engine 同エンジンモデルのパラメータを示す図Diagram showing the parameters of the engine model 同カルマンフィルタの予測（推定）ステップと更新（修正）ステップの概念と計算式を示す図A diagram showing the concept and calculation formula of the prediction (estimation) step and update (correction) step of the Kalman filter. 同カルマンフィルタと因子分析との関係を示す図Diagram showing the relationship between the Kalman filter and factor analysis 同因子スコアによる異常検知の例を示す図Figure showing an example of anomaly detection by the same factor score
　以下に、本発明の実施形態によるエンジンの異常診断方法、エンジンの異常診断プログラム及びエンジンの異常診断システムについて説明する。 The engine abnormality diagnosis method, the engine abnormality diagnosis program, and the engine abnormality diagnosis system according to the embodiment of the present invention will be described below.
　図１はエンジンの異常診断システムのブロック図である。
　エンジンの異常診断システムは、船舶等に搭載されているエンジン１の異常を数学的なエンジンモデル１０を用いて診断する。
　エンジンの異常診断システムは、エンジンモデル１０の初期状態量を入力する条件入力手段２、エンジン１の状態を計測し計測状態量を得る状態量計測手段３、エンジンの異常診断方法又はプログラムを実行するコンピュータ４、コンピュータ４より出力されるエンジン１の異常の診断結果を含む異常情報を提供する情報提供手段（ヒューマンインターフェース手段：ＨＭＩ）５、異常情報の出力に基づいて異常時にエンジン１を制御する異常時制御手段６、情報提供手段５で提供される異常情報を他の箇所に送信する送信手段７、及び各機器をオンラインで接続する接続手段８を備えている。エンジンモデル１０は、エンジン１に対応して最初からコンピュータ４に組み込まれていることが好ましいが、例えば船舶のメンテナンス等に関連してエンジン１の一部又は全部が異なった状態に変更された場合は、条件入力手段２等を介して途中で変更することも可能である。
　接続手段８は、例えばルーターやＬＡＮ等であり、状態量計測手段３と、コンピュータ４と、情報提供手段５とをオンラインで接続する。これにより、エンジン１の異常を含む診断結果を、オンラインでリアルタイムに伝送して提供することができる。なお、オンライン接続は、接続手段８を介して無線、有線いずれも利用することが可能である。 FIG. 1 is a block diagram of an engine abnormality diagnosis system.
The engine abnormality diagnosis system diagnoses an abnormality of the engine 1 mounted on a ship or the like by using a mathematical engine model 10.
The engine abnormality diagnosis system executes a condition input means 2 for inputting an initial state amount of the engine model 10, a state amount measuring means 3 for measuring the state of the engine 1 and obtaining a measured state amount, an engine abnormality diagnosis method or a program. Computer 4, information providing means (human interface means: HMI) 5 that provides abnormality information including the diagnosis result of abnormality of engine 1 output from computer 4, abnormality that controls engine 1 at the time of abnormality based on output of abnormality information The time control means 6, the transmission means 7 for transmitting the abnormality information provided by the information providing means 5 to other places, and the connection means 8 for connecting each device online are provided. It is preferable that the engine model 10 is incorporated in the computer 4 from the beginning corresponding to the engine 1, but when a part or all of the engine 1 is changed to a different state in connection with maintenance of a ship, for example. Can be changed on the way via the condition input means 2 or the like.
The connection means 8 is, for example, a router, a LAN, or the like, and connects the state quantity measuring means 3, the computer 4, and the information providing means 5 online. Thereby, the diagnosis result including the abnormality of the engine 1 can be transmitted and provided online in real time. The online connection can be either wireless or wired via the connection means 8.
　送信手段７はコンピュータ４に設けられている。また、コンピュータ４は、制御部１１、初期状態量取得部１２、エンジン状態推定部１３、計測状態量取得部１４、カルマンフィルタリング部１５、繰返部１６、相関計算部１７、因子分析部１８、異常検知部１９、機械学習適用部２０、異常診断部２１、エンジンモデル更新部２２、主メモリ２３、補助メモリ２４、エンジンモデル活用部２７、出力部２８等を備える。
　補助メモリ２４は、例えばハードディスク等である。補助メモリ２４には、エンジンモデル１０、因子負荷量スペース２５、及び機関故障データベース２６が格納されている。エンジンモデル１０は、エンジン１の仕様及び特性に基づいて予め構築されたものである。因子負荷量スペース２５には、エンジン１の異常を検知してから因子負荷量データが蓄積される。また、機関故障データベース２６には、シミュレーションプログラムでエンジン１の故障のシミュレーションを行い収集したデータが集積される。なお、実際に故障したときのエンジン状態値がある場合は、そのエンジン状態値を機関故障データベース２６に集積することもできる。 The transmission means 7 is provided in the computer 4. Further, the computer 4 includes a control unit 11, an initial state quantity acquisition unit 12, an engine state estimation unit 13, a measurement status quantity acquisition unit 14, a Kalman filtering unit 15, a repeat unit 16, a correlation calculation unit 17, and a factor analysis unit 18. It includes an abnormality detection unit 19, a machine learning application unit 20, an abnormality diagnosis unit 21, an engine model update unit 22, a main memory 23, an auxiliary memory 24, an engine model utilization unit 27, an output unit 28, and the like.
The auxiliary memory 24 is, for example, a hard disk or the like. The engine model 10, the factor load space 25, and the engine failure database 26 are stored in the auxiliary memory 24. The engine model 10 is pre-built based on the specifications and characteristics of the engine 1. In the factor load amount space 25, the factor load amount data is accumulated after detecting the abnormality of the engine 1. Further, in the engine failure database 26, the data collected by simulating the failure of the engine 1 by the simulation program is accumulated. If there is an engine state value at the time of actual failure, the engine state value can be accumulated in the engine failure database 26.
　条件入力手段２は、マウス、キーボード、及びタッチパネル等である。コンピュータ４の操作者、又はユーザ１０１は、条件入力手段２を用いて初期状態量を入力する。初期状態量としては、エンジン１の負荷(Ｑｐ)と燃料ポンプラック位置(ｈｐ)を含む燃料供給量を入力することが好ましい。これにより、エンジン１の状態を推定する上で重要であるエンジン１の負荷(Ｑｐ)と燃料ポンプラック位置(ｈｐ)を含む燃料供給量に基づいて推定状態量を得ることができる。入力された初期状態量は、コンピュータ４の初期状態量取得部１２へ送信され取得される。また、エンジンモデル１０が変わるくらい運転条件が変わったときに、条件入力手段２により、条件信号を入力することもできる。 The condition input means 2 is a mouse, a keyboard, a touch panel, or the like. The operator of the computer 4 or the user 101 inputs the initial state quantity using the condition input means 2. The initial state quantity, it is preferable to enter the fuel supply amount including the load of the engine 1 and the (Q p) fuel pump rack position (h p). Thus, it is possible to obtain the estimated state quantity based on the fuel supply amount including the load of the engine 1 is important for estimating the state of the engine 1 (Q p) and a fuel pump rack position (h p). The input initial state quantity is transmitted to and acquired by the initial state quantity acquisition unit 12 of the computer 4. Further, when the operating conditions change to the extent that the engine model 10 changes, the condition input means 2 can input the condition signal.
　状態量計測手段３は各種センサ等である。状態量計測手段３では、計測状態量としてエンジン１の回転数(ｎｅ)を得ることが好ましい。これにより、エンジン１として計測する機会の多いエンジン１の回転数(ｎｅ)をエンジン１の異常診断に用いることができ、計測精度も高くすることができる。なお、状態量計測手段３で計測される計測状態量としては、この他に例えばエンジン１の掃気圧（Ｐｓ）、排気ガス温度（Ｔｅ）、及びエンジン１の負荷（Ｑｐ）等がある。状態量計測手段３によって計測された計測状態量は、コンピュータ４の計測状態量取得部１４へ送信される。また、計測状態量としては、計測値そのもの以外に、計測値のある期間の平均値や平均値との差など、計測値に基づいて処理したものであってもよい。 The state quantity measuring means 3 is various sensors or the like. In the state quantity measuring means 3, it is preferable to obtain the rotation speed (ne ) of the engine 1 as the measured state quantity. As a result, the rotation speed ( ne ) of the engine 1, which is often measured as the engine 1, can be used for the abnormality diagnosis of the engine 1, and the measurement accuracy can be improved. As the measurement state quantity measured by the state quantity measuring means 3, In addition to this example of the engine 1 scavenging air pressure (P s), the exhaust gas temperature (T e), and the load of the engine 1 (Q p) and the like be. The measured state quantity measured by the state quantity measuring means 3 is transmitted to the measurement status quantity acquisition unit 14 of the computer 4. Further, the measured state quantity may be processed based on the measured value, such as the average value of the measured value for a certain period or the difference from the average value, in addition to the measured value itself.
　出力部２８は、コンピュータ４によるエンジン１の診断結果を、送信手段７、情報提供手段（ヒューマンインターフェース手段）５、及び異常時制御手段６へ出力する。
　なお、情報提供手段５、異常時制御手段６、及び送信手段７は、コンピュータ４の外部に設けることも可能である。
　情報提供手段５は、エンジン１の異常の診断結果として、エンジン１の過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、及び排気ガス温度(Ｔｅ)のうち少なくとも１つの結果を提供することが好ましい。これにより、エンジン１の状態を推定する上で有用な過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、又は排気ガス温度(Ｔｅ)についての診断結果を得ることができる。また、情報提供手段５は、エンジン１の異常の診断結果のみならず異常検知情報を含むあらゆる異常に関連した情報を提供することが可能である。
　送信手段７は、出力された診断結果を、コンピュータ４とは別の場所に設置された機器へ接続手段８を介し有線又は無線ＬＡＮを通じて送信する。これにより、エンジン１から離れた場所においてもリアルタイムにエンジン１の異常の診断結果を含む異常情報を知ることができる。なお、図１では、異常情報を受信する箇所としてブリッジ１００、及び船社等の陸上のユーザ１０１を示している。例えばコンピュータ４が機関室にある場合、情報提供手段５は異常情報を船内ＬＡＮを通じてブリッジ１００へ送信する。
　情報提供手段（ヒューマンインターフェース手段）５は、例えばモニタやスピーカー等である。情報提供手段（ヒューマンインターフェース手段）５からエンジン１の異常情報が提供されることで、乗組員等は異常時に迅速かつ適切に対応することが可能となる。なお、情報提供手段５には、ヒューマンインターフェース手段以外に、異常情報を一時的に蓄える記憶手段や、異常情報を例えばスマートフォンに転送する転送手段等あらゆる情報の提供に関わる手段を含む。
　異常時制御手段６は、異常検知結果又は異常の診断結果に基づいて異常時にエンジン１を自動的に又は乗組員からの操作によって制御する。異常時にエンジン１を制御することにより、エンジン１の異常が悪化したり、故障に至ることを防止できる。なお、異常時制御手段６に提供される情報又は信号は、情報提供手段５に提供される情報の未加工の情報又は信号であってもよい。 The output unit 28 outputs the diagnosis result of the engine 1 by the computer 4 to the transmission means 7, the information providing means (human interface means) 5, and the abnormality control means 6.
The information providing means 5, the abnormality control means 6, and the transmitting means 7 can be provided outside the computer 4.
Information providing means 5, as a diagnostic result of the engine 1 abnormality, the engine 1 supercharger speed (n tc), scavenging pressure (P s), the scavenging temperature (T s), and the exhaust gas temperature (T e) It is preferable to provide at least one of the results. Thus, useful supercharger speed (n tc) in order to estimate the state of the engine 1, scavenging air pressure (P s), the scavenging temperature (T s), or diagnostic results for the exhaust gas temperature (T e) Can be obtained. Further, the information providing means 5 can provide not only the diagnosis result of the abnormality of the engine 1 but also the information related to all the abnormalities including the abnormality detection information.
The transmission means 7 transmits the output diagnosis result to a device installed at a place different from the computer 4 via the connection means 8 via a wired or wireless LAN. As a result, it is possible to know the abnormality information including the diagnosis result of the abnormality of the engine 1 in real time even at a place away from the engine 1. Note that FIG. 1 shows a bridge 100 and a land user 101 such as a shipping company as a location for receiving abnormality information. For example, when the computer 4 is in the engine room, the information providing means 5 transmits the abnormality information to the bridge 100 through the inboard LAN.
The information providing means (human interface means) 5 is, for example, a monitor, a speaker, or the like. By providing the abnormality information of the engine 1 from the information providing means (human interface means) 5, the crew and the like can respond promptly and appropriately in the event of an abnormality. In addition to the human interface means, the information providing means 5 includes a storage means for temporarily storing abnormal information, a transfer means for transferring the abnormal information to a smartphone, and other means related to the provision of all kinds of information.
The abnormality control means 6 controls the engine 1 automatically or by an operation from the crew at the time of abnormality based on the abnormality detection result or the abnormality diagnosis result. By controlling the engine 1 at the time of abnormality, it is possible to prevent the abnormality of the engine 1 from getting worse or leading to a failure. The information or signal provided to the abnormal time control means 6 may be raw information or signal of the information provided to the information providing means 5.
　図２はエンジンの異常診断方法のフローチャート、図３はエンジンの異常診断の概要図である。なお、エンジンの異常診断方法はプログラムとして提供可能であるため、以下では、エンジンの異常診断プログラムがエンジンの異常診断方法における各ステップをコンピュータに実行させるものとして説明する。
　パソコンがエンジンの異常診断プログラムの実行を開始すると、制御部１１は、補助メモリ２４に記憶されているエンジンモデル１０を読み込む（エンジンモデル読込ステップＳ１）。
　エンジンモデル読込ステップＳ１の後、初期状態量取得部１２は、条件入力手段２で入力されたエンジンモデル１０の初期状態量を取得する（初期状態量取得ステップＳ２）
　初期状態量取得ステップＳ２の後、制御部１１は、エンジンモデル活用部２７において読み込んだエンジンモデル１０に初期状態量を適用しエンジンモデル１０を活用する（エンジンモデル活用ステップＳ３）。エンジンモデル活用ステップＳ３においては、エンジンモデル１０の初期設定や更新が行われる。エンジンモデル１０を使うために任意の初期状態量を適用する必要があるが、エンジンモデル１０を活用するエンジンモデル活用ステップＳ３において出来るだけ現在の状態に近い初期状態量（例えば負荷や燃料量）を適用すれば、エンジンモデル１０での計算時間が少なくなったり、エンジンモデル１０による推定計算の精度が上がる等のエンジンモデル１０の活用が図られる。また、カルマンフィルタリングステップＳ６での残差の入力、または取得した計測状態量を処理した結果をエンジンモデル活用ステップＳ３において活用することによりエンジンモデル１０の更新が行われ、エンジンモデル１０が実物のエンジン１に忠実になり活用が図られる。
　エンジン状態推定部１３は、エンジンモデル１０で初期状態量に基づいてエンジン１の状態を計算し推定状態量を得る（エンジン状態推定ステップＳ４）。
　また、計測状態量取得部１４は、状態量計測手段３による計測により得られたエンジン１の計測状態量を取得する（計測状態量取得ステップＳ５）。 FIG. 2 is a flowchart of an engine abnormality diagnosis method, and FIG. 3 is a schematic diagram of engine abnormality diagnosis. Since the engine abnormality diagnosis method can be provided as a program, it will be described below that the engine abnormality diagnosis program causes a computer to execute each step in the engine abnormality diagnosis method.
When the personal computer starts executing the engine abnormality diagnosis program, the control unit 11 reads the engine model 10 stored in the auxiliary memory 24 (engine model reading step S1).
After the engine model reading step S1, the initial state quantity acquisition unit 12 acquires the initial state quantity of the engine model 10 input by the condition input means 2 (initial state quantity acquisition step S2).
After the initial state quantity acquisition step S2, the control unit 11 applies the initial state quantity to the engine model 10 read by the engine model utilization unit 27 and utilizes the engine model 10 (engine model utilization step S3). In the engine model utilization step S3, the engine model 10 is initially set and updated. It is necessary to apply an arbitrary initial state amount in order to use the engine model 10, but in the engine model utilization step S3 that utilizes the engine model 10, the initial state amount (for example, load or fuel amount) that is as close to the current state as possible is applied. If applied, the engine model 10 can be utilized such that the calculation time in the engine model 10 is reduced and the accuracy of the estimation calculation by the engine model 10 is improved. Further, the engine model 10 is updated by inputting the residual in the Kalman filtering step S6 or utilizing the result of processing the acquired measurement state quantity in the engine model utilization step S3, and the engine model 10 is the actual engine. It will be faithful to 1 and will be utilized.
The engine state estimation unit 13 calculates the state of the engine 1 based on the initial state amount in the engine model 10 and obtains the estimated state amount (engine state estimation step S4).
Further, the measurement state quantity acquisition unit 14 acquires the measurement status quantity of the engine 1 obtained by the measurement by the status quantity measuring means 3 (measurement status quantity acquisition step S5).
　カルマンフィルタリング部１５は、取得した計測状態量と計算した推定状態量との残差を非線形カルマンフィルタに入力する（カルマンフィルタリングステップＳ６）。カルマンフィルタリングステップＳ６においては、状態推定とモデルパラメータの推定を行う。
　カルマンフィルタリングステップＳ６の後、エンジンモデル更新部２２は、エンジンモデル１０を更新する必要があるか否かを判定する（モデル更新判定ステップＳ７）。
　エンジンモデル更新部２２は、モデル更新判定ステップＳ７において「Ｙｅｓ」、すなわちエンジンモデル１０を更新する必要があると判定した場合は、カルマンフィルタリングステップＳ６で得られた結果、又は取得した計測状態量を処理した結果をエンジンモデル活用ステップＳ３に適用して、モデルパラメータを更新することによりエンジンモデル１０を更新する（モデル更新ステップＳ８）。エンジンモデル１０を更新することで、エンジン１の経年劣化等に対応し、推定状態量の計算精度を常に高い状態に保つことができる。また、コンピュータ４でエンジンモデル１０を更新することにより、推定状態量の計算精度の向上が容易にできる。モデル更新判定ステップＳ７における判定は、予めモデルパラメータに閾値を設定することにより行われる。
　一方、モデル更新判定ステップＳ７において「Ｎｏ」、すなわちエンジンモデル１０を更新する必要がないとエンジンモデル更新部２２が判定した場合、繰返部１６は、所定の時間が経過したか否かを判定する（時間経過判定ステップＳ９）。繰返部１６は、例えば０．１秒などの時間で区切って、ｋ、ｋ+１、ｋ+２…と繰り返す。
　繰返部１６は、時間経過判定ステップＳ９において「Ｎｏ」、すなわち所定の時間が経過していないと判定した場合は、非線形カルマンフィルタに入力して得られたカルマンゲインをエンジンモデル１０に適用し、エンジン状態推定ステップＳ４と、計測状態量取得ステップＳ５と、カルマンフィルタリングステップＳ６を繰り返す（繰返ステップＳ１０）。
　このように、状態推定とモデルパラメータ推定というカルマンフィルタの二つの機能を使い、エンジンモデル１０の更新と状態推定を繰り返す。なお、計測状態量が確かな場合は、トラッキングフィルタを用いモデルパラメータを同定し、カルマンフィルタは状態推定のみを行うこともできる。
　また、非線形カルマンフィルタは、アンセンテッドカルマンフィルタ又は拡張カルマンフィルタとすることが好ましい。これにより、非線形システムであるエンジン１に対して、カルマンゲインをより適切なものとし推定状態量の計算精度を向上させることができる。 The Kalman filtering unit 15 inputs the residual between the acquired measured state quantity and the calculated estimated state quantity to the non-linear Kalman filter (Kalman filtering step S6). In the Kalman filtering step S6, state estimation and model parameter estimation are performed.
After the Kalman filtering step S6, the engine model update unit 22 determines whether or not the engine model 10 needs to be updated (model update determination step S7).
When the engine model update unit 22 determines "Yes" in the model update determination step S7, that is, it is necessary to update the engine model 10, the result obtained in the Kalman filtering step S6 or the acquired measurement state quantity is used. The processed result is applied to the engine model utilization step S3, and the engine model 10 is updated by updating the model parameters (model update step S8). By updating the engine model 10, it is possible to cope with the aged deterioration of the engine 1 and keep the calculation accuracy of the estimated state quantity always high. Further, by updating the engine model 10 on the computer 4, it is possible to easily improve the calculation accuracy of the estimated state quantity. The determination in the model update determination step S7 is performed by setting a threshold value in the model parameter in advance.
On the other hand, when the engine model update unit 22 determines in the model update determination step S7 that "No", that is, the engine model 10 does not need to be updated, the repeat unit 16 determines whether or not a predetermined time has elapsed. (Time lapse determination step S9). The repeating unit 16 repeats k, k + 1, k + 2, ..., Divided by a time such as 0.1 seconds.
When the repeat unit 16 determines "No" in the time lapse determination step S9, that is, when it is determined that a predetermined time has not elapsed, the repeat unit 16 applies the Kalman gain obtained by inputting to the nonlinear Kalman filter to the engine model 10. The engine state estimation step S4, the measurement state quantity acquisition step S5, and the Kalman filtering step S6 are repeated (repeated step S10).
In this way, the two functions of the Kalman filter, state estimation and model parameter estimation, are used to repeat the update and state estimation of the engine model 10. If the measured state quantity is certain, the tracking filter can be used to identify the model parameters, and the Kalman filter can only estimate the state.
Further, the nonlinear Kalman filter is preferably an unsented Kalman filter or an extended Kalman filter. As a result, the Kalman gain can be made more appropriate for the engine 1 which is a nonlinear system, and the calculation accuracy of the estimated state quantity can be improved.
　一方、時間経過判定ステップＳ９において「Ｙｅｓ」、すなわち所定の時間が経過したと繰返部１６が判定した場合、相関計算部１７は、非線形カルマンフィルタへの入力時に得た残差の相関を計算する（相関計算ステップＳ１１）。相関計算ステップＳ１１における残差の相関の計算は、相関行列に基づいて行うことが好ましい。これにより、簡単な計算により残差の相関の計算精度を向上させることができる。
　なお、非線形カルマンフィルタへの入力時に得た残差の代わりに、取得した計測状態量を用いて、相関を計算することもできる。 On the other hand, when "Yes" in the time lapse determination step S9, that is, when the repeat unit 16 determines that a predetermined time has elapsed, the correlation calculation unit 17 calculates the correlation of the residual obtained at the time of input to the nonlinear Kalman filter. (Correlation calculation step S11). The calculation of the residual correlation in the correlation calculation step S11 is preferably performed based on the correlation matrix. This makes it possible to improve the calculation accuracy of the residual correlation by a simple calculation.
It is also possible to calculate the correlation by using the acquired measurement state quantity instead of the residual obtained at the time of input to the nonlinear Kalman filter.
　相関計算ステップＳ１１の後、因子分析部１８は、残差の相関に対して因子分析し因子負荷量を求める（因子分析ステップＳ１２）。
　なお、残差の相関の代わりに、取得した計測状態量の相関を用いて因子分析を行い、因子負荷量を求めることもできる。
　因子分析ステップＳ１２の後、異常検知部１９は、因子負荷量から因子スコアを計算し異常を検知する（異常検知ステップＳ１３）。異常検知ステップＳ１３で検知された異常は、異常情報として出力することもできる。
　因子分析ステップＳ１２においては、相関行列としての共分散行列に基づいて特異値分解(ＳＶＤ)をして因子負荷量を導出することが好ましい。これにより、共分散行列の形に捉われず本質的に重要なものを抽出し、エンジン１の異常を早期に検知することができる。 After the correlation calculation step S11, the factor analysis unit 18 performs factor analysis on the correlation of the residuals to obtain the factor loading (factor analysis step S12).
Instead of the correlation of the residual, the factor analysis can be performed using the correlation of the acquired measurement state quantity to obtain the factor loading amount.
After the factor analysis step S12, the abnormality detection unit 19 calculates the factor score from the factor loading amount and detects the abnormality (abnormality detection step S13). The abnormality detected in the abnormality detection step S13 can also be output as abnormality information.
In the factor analysis step S12, it is preferable to derive the factor loading by performing singular value decomposition (SVD) based on the covariance matrix as the correlation matrix. As a result, it is possible to extract what is essentially important regardless of the shape of the covariance matrix and detect the abnormality of the engine 1 at an early stage.
　因子分析ステップＳ１２で求めた因子負荷量は因子負荷量スペース２５に蓄積される（因子負荷量蓄積ステップＳ１４）。機械学習適用部２０は、機関故障データベース２６に記憶されているデータを読み出す（機関故障データ読出ステップＳ１５）と共に、因子負荷量を機械学習に適用する（機械学習適用ステップＳ１６）。機械学習は、自己組織化マップ(ＳＯＭ)を用いる。これにより、教師なし機械学習であるＳＯＭを利用してエンジン１の異常の原因を明確に分類することができる。なお、機械学習アルゴリズムとしては、ＳＯＭ以外にＳＶＭ（Support vector machine）や、Fuzzy C-means等を用いることもできる。
　異常診断部２１は、機械学習に基づいて異常を診断する（異常診断ステップＳ１７）。
　異常検知ステップＳ１３及び異常診断ステップＳ１７の後、出力部２８は、エンジン１の異常の診断結果を含む異常情報を出力する（出力ステップＳ１８）。出力ステップＳ１８における異常情報の出力には異常の診断結果以外にも異常検知情報や付随した情報を含めることができる。出力先は、上述のように情報提供手段（ヒューマンインターフェース手段）５、異常時制御手段６、及び送信手段７である。 The factor loading amount obtained in the factor analysis step S12 is accumulated in the factor loading amount space 25 (factor loading amount accumulation step S14). The machine learning application unit 20 reads the data stored in the engine failure database 26 (engine failure data read step S15) and applies the factor load to machine learning (machine learning application step S16). Machine learning uses a self-organizing map (SOM). Thereby, the cause of the abnormality of the engine 1 can be clearly classified by using SOM which is unsupervised machine learning. As the machine learning algorithm, SVM (Support vector machine), Fuzzy C-means, or the like can be used in addition to SOM.
The abnormality diagnosis unit 21 diagnoses an abnormality based on machine learning (abnormality diagnosis step S17).
After the abnormality detection step S13 and the abnormality diagnosis step S17, the output unit 28 outputs the abnormality information including the abnormality diagnosis result of the engine 1 (output step S18). In the output of the abnormality information in the output step S18, the abnormality detection information and the accompanying information can be included in addition to the abnormality diagnosis result. As described above, the output destinations are the information providing means (human interface means) 5, the abnormality control means 6, and the transmitting means 7.
　以下に因子分析を用いたエンジン１の異常診断について詳細に説明する。
　図４は因子分析の概念図である。因子分析は、すべての計測値（計測状態量）ｙｍに共通で、その関係性がａｉｍで表せる隠れた（計測できない）変数因子Ｆを探すことである。これは線形の関係パラメータａを使って下式（１）のように表される。
Figure JPOXMLDOC01-appb-M000001
 　式１において、ｆは共通因子、ａｉｍは線形係数、残ったｕｉは説明できない因子で計測エラーかノイズである。
　個々の計測値ｙｍは、ある数の共通因子ｆに線形で結びついている。線形係数ａｉｍは因子負荷量と呼ばれる。因子分析とは隠れた変数（因子）を探すことであるが、この因子は計測値が変化した際の何らかの異常（事故原因）とみなすことができる。
　図５は計測データを用いた因子分析の例を示す図である。図５に示すように、計測データＹを使った因子分析の例として、計測値の具体的な追跡変化による主因子と因子負荷量Ａの計算例を挙げると、計測値ｙ１，ｙ２，.....ｙｍは例えばエンジン１の掃気圧，排ガス温度，.....エンジン負荷を表し、ある時刻の区間（例えば０～ｔ区間）をひとまとめにして行列を作る（図５（ａ））。この行列の共分散行列を計算し各変数の標準偏差で割ってＲを求める（図５（ｂ））。Ｒを特異値分解（ＳＶＤ）して特異値Ｓを求める。次に第１因子負荷量Ａを求める（図５（ｃ））。各第１因子負荷量ａの二乗を足して分散の総和で割ったものが因子スコアＤ（インデックス）である（図５（ｄ））。
　なお、共分散行列の相関行列を特異値分解（ＳＶＤ）することにより第１段階の因子負荷量を求め、これをＥＭ（Expectation Maximization)法を使い、さらに特徴のある顕著な因子負荷量を求めることもできる。 The abnormality diagnosis of the engine 1 using the factor analysis will be described in detail below.
FIG. 4 is a conceptual diagram of factor analysis. Factor analysis is common to all the measurement values (measurement state amount) y m, is to look for the relationship is hidden expressed by a im (can not be measured) variable factor F. This is expressed by the following equation (1) using the linear relation parameter a.
Figure JPOXMLDOC01-appb-M000001
 In Equation 1, f is common factor, a im is linear coefficients, remaining u i is the measurement error or noise factors that can not be explained.
Individual measurements y m is linked in a linear common factor f of a number. The linear coefficient a im is called a factor loading. Factor analysis is to search for hidden variables (factors), and this factor can be regarded as some abnormality (accident cause) when the measured value changes.
FIG. 5 is a diagram showing an example of factor analysis using measurement data. As shown in FIG. 5, as an example of factor analysis using the measurement data Y, the calculation examples of the main factors and factor loadings A according to a specific track changes in the measured value, the measured value y 1, y 2, ..... y m for example scavenging air pressure of the engine 1, exhaust gas temperature, represents ..... engine load, create a matrix collectively a section of a certain time (e.g., 0 ~ t period) (Fig. 5 ( a)). The covariance matrix of this matrix is calculated and divided by the standard deviation of each variable to obtain R (FIG. 5 (b)). R is decomposed into singular values (SVD) to obtain the singular value S. Next, the first factor loading amount A is obtained (FIG. 5 (c)). The factor score D (index) is obtained by adding the squares of each first factor loading amount a and dividing by the sum of the variances (FIG. 5 (d)).
By singular value decomposition (SVD) of the correlation matrix of the covariance matrix, the factor loading of the first stage is obtained, and this is obtained by using the EM (Expectation Maximization) method to obtain a more characteristic and remarkable factor loading. You can also do it.
　図６は因子スコアの例を示す図であり、図６（ａ）は生の因子スコアＦ１（Ｄ１）を示し、図６（ｂ）はフィルタ後の因子スコアＦ１（Ｄ１）を示している。また、図７は各計測値の因子負荷量の変化を示す図であり、図７（ａ）は掃気圧、図７（ｂ）は過給機回転数、図７（ｃ）は排気ガス温度、図７（ｄ）は掃気温度である。因子スコアＤ１（主インデックス）は、時系列に並べていくと、例えば図６に示すようにエンジン１の空気冷却器の冷却水の流量が減り始めた時など、エンジン１に異常（何らかの変化）が起こった時に変化がみられる。
　換言すると、図７に示すように因子負荷量行列Ａに含まれている情報は推進システムの何らかの変化による異常を表すものである。更に、因子負荷量（行列Ａの行）は誤差の要素としての推進システムパラメータ間の関係強度を表しており、異常原因の特徴を表す。従って、因子負荷量を機械学習アルゴリズム、例えば自己組織化マップ（ＳＯＭ）等を使い機械学習をすることにより、推進システムの事故の原因分類に利用される。また、因子スコアＤ１は異常の早期検知に利用される。 6A and 6B are diagrams showing an example of factor scores, FIG. 6A shows a raw factor score F1 (D1), and FIG. 6B shows a filtered factor score F1 (D1). Further, FIG. 7 is a diagram showing changes in the factor load of each measured value, FIG. 7 (a) is a scavenging pressure, FIG. 7 (b) is a turbocharger rotation speed, and FIG. 7 (c) is an exhaust gas temperature. , FIG. 7 (d) is a scavenging temperature. When the factor scores D1 (main index) are arranged in chronological order, there is an abnormality (some change) in the engine 1 such as when the flow rate of the cooling water of the air cooler of the engine 1 starts to decrease as shown in FIG. Changes are seen when it happens.
In other words, as shown in FIG. 7, the information contained in the factor loading matrix A represents an abnormality due to some change in the propulsion system. Further, the factor loading (row of the matrix A) represents the strength of the relationship between the propulsion system parameters as an element of error, and represents the characteristics of the cause of the abnormality. Therefore, the factor load is machine-learned using a machine learning algorithm, for example, a self-organizing map (SOM), etc., and is used for classifying the cause of an accident in the propulsion system. In addition, the factor score D1 is used for early detection of abnormalities.
　図３に示すように本発明ではカルマンフィルタリング部１５としてカルマンフィルタ観測器を用い、ここでは、計測状態量と推定状態量との残差を使い因子分析をする。すなわち本発明では、もう一つの方法として計測値（計測状態量）Ｙの代わりに計測状態量とエンジン状態推定部１３による推定状態量とのカルマンフィルタリング部１５で固有に計算される残差Ｅを用いて上記と同様の因子分析を行う。この残差Ｅは推定状態量からの乖離を意味するものであり、推定状態量が正常状態と考えると正常状態からの乖離を反映するものである。したがって、エンジン１に何らかの異常が発生したことを残差Ｅに基づいて検知できる。
　カルマンフィルタ観測器は、デジタルツインのエンジンモデル１０をベースとしている。カルマンフィルタ観測器においては、エンジン１の動的プロセスにより取得した計測状態量と、初期状態量に基づき数学的なエンジンモデル１０のプロセスにより計算した推定状態量との残差Ｅを非線形カルマンフィルタに入力する。これによりカルマンゲインが得られる。カルマンゲインは、エンジンモデル１０に適用され、数学的なエンジンモデル１０のプロセス制御に用いられる。
　このように、エンジン１のデジタルツインモデルとしてのエンジンモデル１０を用いて、エンジン状態をモニタリングすることにより、早い段階でエンジン１の故障を検知し、原因を診断することができる。 As shown in FIG. 3, in the present invention, a Kalman filter observer is used as the Kalman filtering unit 15, and here, factor analysis is performed using the residual between the measured state quantity and the estimated state quantity. That is, in the present invention, as another method, instead of the measured value (measured state amount) Y, the residual E uniquely calculated by the Kalman filtering unit 15 between the measured state amount and the estimated state amount by the engine state estimation unit 13 is used. Perform the same factor analysis as above. This residual E means a deviation from the estimated state quantity, and when the estimated state quantity is considered to be a normal state, it reflects the deviation from the normal state. Therefore, it is possible to detect that some abnormality has occurred in the engine 1 based on the residual E.
The Kalman filter observer is based on the digital twin engine model 10. In the Kalman filter observer, the residual E between the measured state quantity acquired by the dynamic process of the engine 1 and the estimated state quantity calculated by the process of the mathematical engine model 10 based on the initial state quantity is input to the nonlinear Kalman filter. .. As a result, Kalman gain is obtained. The Kalman gain is applied to the engine model 10 and is used for the process control of the mathematical engine model 10.
In this way, by monitoring the engine state using the engine model 10 as the digital twin model of the engine 1, it is possible to detect the failure of the engine 1 at an early stage and diagnose the cause.
　図８はエンジンの数学モデルの例を示す図である。
　図８中の矢印の左側は、実際のエンジン１における燃料供給系（燃料ポンプラック位置（ｈｐ））とエンジン１の回転数（ｎｅ）等の計測系の関係図であり、エンジン１における計測点と計測値を示している。図８中の右側は、エンジン１の数学モデルであるエンジンモデル１０を示している。
　また、図８においては状態量計測手段３を〇囲みの英文字で示している。〇で囲った「Ｔ」は温度計、〇で囲った「Ｐ」は圧力計、〇で囲った「ｎ」は回転数計、〇で囲った「Ｑ」は軸馬力計である。負荷変動は軸馬力計で計測する。
　計測値は毎ステップ（ｋ，ｋ＋１，ｋ＋２，・・・）計測し、エンジンモデル１０で毎ステップ計算（推定）する。エンジンモデル１０の推定精度をあげるために、もっとも確実に高い精度で計測できるエンジン１の回転数（ｎｅ）を取得してカルマンゲインを算出してエンジンモデル１０を修正していく。これがカルマンフィルタリングである。
　図８に示すエンジン１の数学モデルにおいて、通常状態における挙動は非線形の状態空間モデルで表すことができる。非線形の状態空間モデルは、状態方程式Ｘで表される方程式、状態量ｘで表される各パラメータ、入力量ｕで表される入力で構成される。状態方程式Ｘの右辺は、右側の真ん中のブロックに示される各関数で表現され、状態量ｎｅ,ｎｔｃ,Ｐｓ,Ｔｅ,Ｐｅ,Ｇｆと入力量ｈｐ、Ｑｐとの関係が５つの式で表現される。
例えば、入力として燃料ポンプラック位置（ｈｐ）とエンジンの負荷（Ｑｐ）をとり、状態（ｘ）と出力（ｙ）として、エンジン１の回転数（ｎｅ）、過給機回転数（ｎｔｃ）、掃気圧（Ｐｓ）、排気ガス圧（Ｐｅ）、排気ガス温度（Ｔｅ）、燃料流量（Ｇｆ）をとり、それぞれのシステム関数Ｆと観測方程式ｙ（ｔ）の出力関数Ｈを表す。
図８の例では出力される観測値ｙは状態量ｘと等しい。 FIG. 8 is a diagram showing an example of a mathematical model of an engine.
Left arrow in Figure 8, the actual fuel supply system in the engine 1 (fuel pump rack position (h p)) and a relational diagram of the measurement system, such as the engine revolution speed (n e), the engine 1 The measurement points and measurement values are shown. The right side in FIG. 8 shows an engine model 10 which is a mathematical model of the engine 1.
Further, in FIG. 8, the state quantity measuring means 3 is shown by the enclosing English characters. "T" surrounded by 〇 is a thermometer, "P" surrounded by 〇 is a pressure gauge, "n" surrounded by 〇 is a tachometer, and "Q" surrounded by 〇 is a shaft horsepower meter. Load fluctuations are measured with an axial horsepower meter.
The measured value is measured at each step (k, k + 1, k + 2, ...), And is calculated (estimated) at each step by the engine model 10. In order to improve the estimation accuracy of the engine model 10, the rotation speed ( ne ) of the engine 1 that can be measured with the highest accuracy is acquired, the Kalman gain is calculated, and the engine model 10 is modified. This is Kalman filtering.
In the mathematical model of the engine 1 shown in FIG. 8, the behavior in the normal state can be represented by a non-linear state-space model. The non-linear state space model is composed of an equation represented by the equation of state X, each parameter represented by the state quantity x, and an input represented by the input quantity u. State the right-hand side of the equation X is expressed by the function shown in block in the middle right of the state quantities n e, n tc, P s , T e, P e, G f and the input amount h p, and Q p The relationship is expressed by five equations.
For example, taking the fuel pump rack position (h p) and the engine load (Q p) as an input, as a state (x) and the output (y), the engine revolution speed (n e), the supercharger rotational speed ( n ct ), sweep pressure (P s ), exhaust gas pressure (P e ), exhaust gas temperature (T e ), fuel flow rate (G f ), and output of each system function F and observation equation y (t). Represents the function H.
In the example of FIG. 8, the output observed value y is equal to the state quantity x.
　図９はエンジンモデルのパラメータを示す図である。
　モデルパラメータとしては、プロペラトルク等のエンジンの負荷（Ｑｐ）、エンジントルク（Ｑｅ）、慣性モーメント（Ｉｅ，Ｉｔｃ）、燃料ポンプラック位置(ｈｐ)、エンジンの回転数（ｎｅ）、過給機回転数（ｎｔｃ）、大気圧（Ｐａ）、大気温度（Ｔａ）、掃気圧（Ｐｓ）、掃気温度（Ｔｓ）、シリンダ内最大圧縮圧（Ｐｃ）、シリンダ内最大燃焼圧（Ｐｚ）、シリンダ内平均有効圧（Ｐｉ）、掃気レシーバー体積（Ｖａ．ｒ）、排気レシーバー体積（Ｖｅ．ｒ）、熱力学定数（Ｒａ、Ｒｅ、ｋｅ、Ｃｐｅ、Ｃｐａ）、冷却水温度（Ｔｗ）、コンプレッサー出口温度（Ｔｃ）、排気ガス圧（Ｐｅ）、排気ガス温度（Ｔｅ）、タービン出口温度（Ｔｏｕｔ）、タービン出口圧（Ｐｏｕｔ）、燃料流量（Ｇｆ）、掃気流量（Ｇａ）、コンプレッサー空気流量（Ｇｃ）、排気ガス流量（Ｇｅ）が挙げられる。 FIG. 9 is a diagram showing parameters of the engine model.
The model parameters, the load of the engine, such as a propeller torque (Q p), engine torque (Q e), the moment of inertia (I e, I tc), the fuel pump rack position (h p), the rotational speed of the engine (n e ), Supercharger rotation speed ( ntc ), atmospheric pressure (P a ), atmospheric temperature (T a ), scavenging pressure (P s ), scavenging temperature (T s ), maximum compression pressure in the cylinder (P c ), Maximum combustion pressure in cylinder (P z ), average effective pressure in cylinder (P i ), scavenging receiver volume (V a r ), exhaust receiver volume (V e r ), thermodynamic constants (R a , R e , k e, C pe, C pa ), the cooling water temperature (T w), the compressor outlet temperature (T c), the exhaust gas pressure (P e), the exhaust gas temperature (T e), a turbine outlet temperature (T out), turbine outlet pressure (P out), the fuel flow rate (G f), scavenging flow (G a), a compressor air flow rate (G c), include an exhaust gas flow rate (G e).
　図１０はカルマンフィルタの予測（推定）ステップ（カルマンフィルタリングステップＳ６）と更新（修正）ステップ（モデル更新ステップＳ８）の概念と計算式を示す図である。
　エンジンモデル１０の実体との違いと、状態量計測手段３による計測の不確かさを考慮すると、エンジンモデル１０による推定状態量と計測状態量との間に誤差（残差）が生じる。この誤差を計算し、計測状態量で修正しながら、推定状態量をできるだけ正しい値に近づけるのが非線形カルマンフィルタであるアンセンテッドカルマンフィルタである。図１０にアンセンテッドカルマンフィルタの基礎を示す。
　カルマンフィルタを適用することで、離散的な各サンプリング時間ｋにおいて、計測とモデリングの不確かさを考慮に入れながら、推進システムの挙動を計算された誤差で繰り返し表される（下式２）。
Figure JPOXMLDOC01-appb-M000002
 　通常状態の運転では、誤差の分布は０平均の正規分布とみなされる。異常状態を検知するために、カルマンフィルタで生成された誤差を因子分析に供する。 FIG. 10 is a diagram showing the concept and calculation formula of the Kalman filter prediction (estimation) step (Kalman filtering step S6) and the update (correction) step (model update step S8).
Considering the difference from the substance of the engine model 10 and the uncertainty of measurement by the state quantity measuring means 3, an error (residual) occurs between the estimated state quantity by the engine model 10 and the measured state quantity. The unsented Kalman filter, which is a nonlinear Kalman filter, calculates this error and corrects it with the measured state quantity to bring the estimated state quantity as close to the correct value as possible. FIG. 10 shows the basics of the unsented Kalman filter.
By applying the Kalman filter, the behavior of the propulsion system is repeatedly expressed by the calculated error at each discrete sampling time k, taking into account the uncertainty of measurement and modeling (Equation 2 below).
Figure JPOXMLDOC01-appb-M000002
 In normal operation, the error distribution is considered to be a 0 average normal distribution. In order to detect the abnormal state, the error generated by the Kalman filter is used for factor analysis.
　図１１はカルマンフィルタと因子分析との関係を示す図である。
　因子分析モデル（Ｙ＝ＡＦ＋Ｕ）のパラメータＡは、相関行列（共分散行列）から推定される。上述のように、カルマンフィルタ観測器は、デジタルツインのエンジンモデル１０をベースとし、共分散推定が毎回行われ、カルマンゲインはどの状態量を修正すべきかのインジケータとして使われる。 FIG. 11 is a diagram showing the relationship between the Kalman filter and factor analysis.
Parameter A of the factor analysis model (Y = AF + U) is estimated from the correlation matrix (covariance matrix). As mentioned above, the Kalman filter observer is based on the engine model 10 of the digital twin, the covariance estimation is performed every time, and the Kalman gain is used as an indicator of which state quantity should be corrected.
　図１２は因子スコアによる異常検知の例を示す図である。
　図１２は、実エンジン１の過給機吸い込みフィルターの閉塞模擬実験を行い、エンジンの異常診断システムで異常を迅速に検知した例である。なお、グラフの横軸はサンプルの回数であり１回数は０．１秒である。過給機吸い込みフィルターが徐々に閉塞し圧損が増えていくが、圧損が増え始める初期に因子スコアＦ１が急上昇するため異常を検知できる。 FIG. 12 is a diagram showing an example of abnormality detection based on a factor score.
FIG. 12 shows an example in which a blockage simulation experiment of the supercharger suction filter of the actual engine 1 was performed and an abnormality was quickly detected by the engine abnormality diagnosis system. The horizontal axis of the graph is the number of samples, and one time is 0.1 seconds. The turbocharger suction filter is gradually blocked and the pressure loss increases, but the factor score F1 rises sharply at the initial stage when the pressure loss starts to increase, so that an abnormality can be detected.
　このように、計測状態量と計算した推定状態量との残差を利用して因子分析にかけることにより、計算した因子スコアに基づきエンジン１の異常を早期に検知することができる。また、因子負荷量を機械学習に適用しエンジン１の原因を診断することができる。また、コンピュータ４を利用して、エンジン１の異常の早期検知とその原因の診断を行った結果を含む異常情報を提供することができる。
　なお、コンピュータ４の各構成要素及び周辺手段は、適宜、外付けすることや内蔵することが可能であり、コンピュータ４を複数のコンピュータで役割分担をしたり、一部をディスクリート回路とすることも可能である。 In this way, by performing factor analysis using the residual between the measured state quantity and the calculated estimated state quantity, it is possible to detect the abnormality of the engine 1 at an early stage based on the calculated factor score. Further, the factor load can be applied to machine learning to diagnose the cause of the engine 1. Further, the computer 4 can be used to provide abnormality information including the result of early detection of the abnormality of the engine 1 and diagnosis of the cause thereof.
Each component and peripheral means of the computer 4 can be externally attached or incorporated as appropriate, and the computer 4 may be divided into roles among a plurality of computers, or a part of the computer 4 may be a discrete circuit. It is possible.
　本発明は、例えば就航船のエンジンの異常の早期検知と診断を行い、離れた場所においてもリアルタイムにエンジンの異常を含む診断結果を知ることができるため、安全かつ効率的な運航に寄与する。また、船舶以外のエンジンの異常の早期検知と異常診断にも利用することができる。 The present invention contributes to safe and efficient operation because, for example, early detection and diagnosis of an abnormality in the engine of a vessel in service can be performed, and the diagnosis result including the abnormality of the engine can be known in real time even at a remote location. It can also be used for early detection and diagnosis of abnormalities in engines other than ships.
１　エンジン
２　条件入力手段
３　状態量計測手段
４　コンピュータ
５　情報提供手段（ヒューマンインターフェース手段）
６　異常時制御手段
７　送信手段
８　接続手段
１０　エンジンモデル
Ｓ２　初期状態量取得ステップ
Ｓ３　エンジンモデル活用ステップ
Ｓ４　エンジン状態推定ステップ
Ｓ５　計測状態量取得ステップ
Ｓ６　カルマンフィルタリングステップ
Ｓ８　モデル更新ステップ
Ｓ１０　繰返ステップ
Ｓ１１　相関計算ステップ
Ｓ１２　因子分析ステップ
Ｓ１３　異常検知ステップ
Ｓ１６　機械学習適用ステップ
Ｓ１７　異常診断ステップ
Ｓ１８　出力ステップ 1 Engine 2 Condition input means 3 State quantity measuring means 4 Computer 5 Information providing means (human interface means)
6 Abnormal time control means 7 Transmission means 8 Connection means 10 Engine model S2 Initial state quantity acquisition step S3 Engine model utilization step S4 Engine status estimation step S5 Measurement status quantity acquisition step S6 Kalman filtering step S8 Model update step S10 Repeat step S11 Correlation Calculation step S12 Factor analysis step S13 Abnormality detection step S16 Machine learning application step S17 Abnormality diagnosis step S18 Output step

Claims (17)

  1. 　エンジンの異常を、数学的なエンジンモデルを用いて異常診断する方法であって、
    前記エンジンモデルの初期状態量を取得する初期状態量取得ステップと、
    前記エンジンモデルに前記初期状態量を適用し前記エンジンモデルを活用するエンジンモデル活用ステップと、
    前記エンジンモデルで前記初期状態量に基づいて前記エンジンの状態を計算し推定状態量を得るエンジン状態推定ステップと、
    前記エンジンの計測状態量を取得する計測状態量取得ステップと、
    取得した前記計測状態量と計算した前記推定状態量との残差を非線形カルマンフィルタにかけるカルマンフィルタリングステップと、
    前記非線形カルマンフィルタにかけて得られたカルマンゲインを前記エンジンモデルに適用し、前記エンジン状態推定ステップと、前記計測状態量取得ステップと、前記カルマンフィルタリングステップを繰り返す繰返ステップと、
    前記非線形カルマンフィルタにかけたときの前記計測状態量、又は前記残差の相関を計算する相関計算ステップと、
    前記計測状態量、又は前記残差の前記相関に対して因子分析し因子負荷量を求める因子分析ステップと、
    前記因子負荷量から因子スコアを計算し前記異常を検知する異常検知ステップと、
    前記因子負荷量を機械学習に適用する機械学習適用ステップと、
    前記機械学習に基づいて前記異常を診断する異常診断ステップと、
    前記エンジンの前記異常の診断結果を含む異常情報を出力する出力ステップと
    を実行することを特徴とするエンジンの異常診断方法。     It is a method of diagnosing an abnormality of an engine using a mathematical engine model.
    The initial state quantity acquisition step for acquiring the initial state quantity of the engine model, and
    An engine model utilization step that applies the initial state quantity to the engine model and utilizes the engine model,
    An engine state estimation step of calculating the state of the engine based on the initial state amount in the engine model and obtaining an estimated state amount, and
    The measurement state quantity acquisition step for acquiring the measurement status quantity of the engine, and
    A Kalman filtering step that applies a non-linear Kalman filter to the residual between the acquired measured state quantity and the calculated estimated state quantity.
    The Kalman gain obtained by applying the non-linear Kalman filter is applied to the engine model, and the engine state estimation step, the measurement state quantity acquisition step, and the Kalman filtering step are repeated.
    A correlation calculation step for calculating the correlation of the measured state quantity or the residual when applied to the nonlinear Kalman filter.
    A factor analysis step of factor-analyzing the correlation of the measured state quantity or the residual to obtain the factor loading.
    An abnormality detection step that calculates the factor score from the factor loading and detects the abnormality,
    A machine learning application step that applies the factor loading to machine learning,
    An abnormality diagnosis step for diagnosing the abnormality based on the machine learning,
    A method for diagnosing an abnormality of an engine, which comprises executing an output step of outputting abnormality information including the diagnosis result of the abnormality of the engine.
  2. 　前記カルマンフィルタリングステップで得られた結果、又は取得した前記計測状態量を処理した結果を前記エンジンモデル活用ステップに適用して、前記エンジンモデルを更新するモデル更新ステップをさらに実行することを特徴とする請求項１に記載のエンジンの異常診断方法。 It is characterized in that the result obtained in the Kalman filtering step or the result of processing the acquired measurement state quantity is applied to the engine model utilization step, and the model update step for updating the engine model is further executed. The engine abnormality diagnosis method according to claim 1.
  3. 　前記相関計算ステップにおける前記計測状態量、又は前記残差の前記相関の計算は、相関行列に基づいて行うことを特徴とする請求項１又は請求項２に記載のエンジンの異常診断方法。 The engine abnormality diagnosis method according to claim 1 or 2, wherein the calculation of the correlation of the measured state quantity or the residual in the correlation calculation step is performed based on a correlation matrix.
  4. 　前記因子分析ステップにおいて、前記相関行列としての共分散行列に基づいて特異値分解(ＳＶＤ)をして前記因子負荷量を導出することを特徴とする請求項３に記載のエンジンの異常診断方法。 The method for diagnosing an abnormality of an engine according to claim 3, wherein in the factor analysis step, the factor loading is derived by performing singular value decomposition (SVD) based on the covariance matrix as the correlation matrix.
  5. 　前記機械学習は、自己組織化マップ(ＳＯＭ)を用いることを特徴とする請求項１から請求項４のいずれか１項に記載のエンジンの異常診断方法。 The engine abnormality diagnosis method according to any one of claims 1 to 4, wherein the machine learning uses a self-organizing map (SOM).
  6. 　前記初期状態量取得ステップで取得する前記初期状態量は、前記エンジンの負荷(Ｑｐ)と燃料ポンプラック位置(ｈｐ)を含む燃料供給量であることを特徴とする請求項１から請求項５のいずれか１項に記載のエンジンの異常診断方法。 Wherein said initial state quantity acquired in the initial state quantity acquisition step, claim from claim 1, characterized in that the fuel supply amount including load (Q p) and a fuel pump rack position (h p) of the engine The engine abnormality diagnosis method according to any one of 5.
  7. 　前記計測状態量取得ステップで取得する前記計測状態量は、前記エンジンの回転数(ｎｅ)であることを特徴とする請求項１から請求項６のいずれか１項に記載のエンジンの異常診断方法。 The abnormality diagnosis of the engine according to any one of claims 1 to 6, wherein the measured state quantity acquired in the measured state quantity acquisition step is the rotation speed ( ne) of the engine. Method.
  8. 　前記エンジン状態推定ステップの前記推定状態量として、過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、及び排気ガス温度(Ｔｅ)を得ることを特徴とする請求項１から請求項７のいずれか１項に記載のエンジンの異常診断方法。 As the estimated state quantity of the engine state estimating step, the supercharger rotational speed (n tc), scavenging pressure (P s), the scavenging temperature (T s), and the feature to obtain the exhaust gas temperature (T e) The engine abnormality diagnosis method according to any one of claims 1 to 7.
  9. 　エンジンの異常を、数学的なエンジンモデルを用いて異常診断するプログラムであって、
    コンピュータに、請求項１から請求項８のいずれか１項に記載のエンジンの異常診断方法における前記初期状態量取得ステップ、前記エンジンモデル活用ステップ、前記エンジン状態推定ステップ、前記計測状態量取得ステップ、前記カルマンフィルタリングステップ、前記繰返ステップ、前記相関計算ステップ、前記因子分析ステップ、前記異常検知ステップ、前記機械学習適用ステップ、前記異常診断ステップ、及び前記出力ステップを実行させることを特徴とするエンジンの異常診断プログラム。     A program that diagnoses engine abnormalities using a mathematical engine model.
    The initial state amount acquisition step, the engine model utilization step, the engine state estimation step, and the measurement state amount acquisition step in the engine abnormality diagnosis method according to any one of claims 1 to 8 are applied to a computer. An engine characterized in that the Kalman filtering step, the repeat step, the correlation calculation step, the factor analysis step, the abnormality detection step, the machine learning application step, the abnormality diagnosis step, and the output step are executed. Abnormality diagnosis program.
  10. 　エンジンと、
    エンジンモデルの初期状態量を入力する条件入力手段と、
    前記エンジンの状態を計測し計測状態量を得る状態量計測手段と、
    請求項１から請求項８のいずれか１項に記載のエンジンの異常診断方法、又は請求項９に記載のエンジンの異常診断プログラムを実行するコンピュータと、
    前記コンピュータより出力される前記エンジンの異常の診断結果を含む異常情報を提供する情報提供手段とを備えたことを特徴とするエンジンの異常診断システム。     With the engine
    Condition input means for inputting the initial state quantity of the engine model,
    A state quantity measuring means that measures the state of the engine and obtains the measured state quantity,
    A computer that executes the engine abnormality diagnosis method according to any one of claims 1 to 8 or the engine abnormality diagnosis program according to claim 9.
    An engine abnormality diagnosis system including an information providing means for providing abnormality information including an abnormality diagnosis result of the engine output from the computer.
  11. 　前記コンピュータで、前記エンジンモデルの更新を行うことを特徴とする請求項１０に記載のエンジンの異常診断システム。 The engine abnormality diagnosis system according to claim 10, wherein the computer updates the engine model.
  12. 　前記状態量計測手段で、前記計測状態量として前記エンジンの回転数(ｎｅ)を得ることを特徴とする請求項１０又は請求項１１に記載のエンジンの異常診断システム。 The engine abnormality diagnosis system according to claim 10 or 11, wherein the state quantity measuring means obtains the rotation speed ( ne) of the engine as the measured state quantity.
  13. 　前記情報提供手段で、前記異常の診断結果として、前記エンジンの過給機回転数(ｎｔｃ)、掃気圧(Ｐｓ)、掃気温度（Ｔｓ）、及び排気ガス温度(Ｔｅ)の少なくとも１つの結果を提供することを特徴とする請求項１０から請求項１２のいずれか１項に記載のエンジンの異常診断システム。 In the information providing unit, as a diagnosis result of the abnormality, the supercharger rotation speed of the engine (n tc), scavenging pressure (P s), the scavenging temperature (T s), and at least the exhaust gas temperature (T e) The engine abnormality diagnosis system according to any one of claims 10 to 12, wherein one result is provided.
  14. 　前記異常情報の前記出力に基づいて、異常時に前記エンジンを制御する異常時制御手段を備えたことを特徴とする請求項１０から請求項１３のいずれか１項に記載のエンジンの異常診断システム。 The engine abnormality diagnosis system according to any one of claims 10 to 13, further comprising an abnormality control means for controlling the engine at the time of abnormality based on the output of the abnormality information.
  15. 　前記情報提供手段として、ヒューマンインターフェース手段を用いて前記異常情報を提供することを特徴とする請求項１０から請求項１４のいずれか１項に記載のエンジンの異常診断システム。 The engine abnormality diagnosis system according to any one of claims 10 to 14, wherein the abnormality information is provided by using a human interface means as the information providing means.
  16. 　前記情報提供手段で提供される前記異常情報を、他の箇所に送信する送信手段を備えたことを特徴とする請求項１０から請求項１５のいずれか１項に記載のエンジンの異常診断システム。 The engine abnormality diagnosis system according to any one of claims 10 to 15, further comprising a transmission means for transmitting the abnormality information provided by the information providing means to another location.
  17. 　前記状態量計測手段と、前記コンピュータと、前記情報提供手段とをオンラインで接続する接続手段を備えたことを特徴とする請求項１０から請求項１６のいずれか１項に記載のエンジンの異常診断システム。 The abnormality diagnosis of the engine according to any one of claims 10 to 16, further comprising a connection means for connecting the state quantity measuring means, the computer, and the information providing means online. system.
PCT/JP2021/025526 2020-07-06 2021-07-06 Engine abnormality diagnosis method, engine abnormality diagnosis program, and engine abnormality diagnosis system WO2022009904A1 (en)

Priority Applications (3)

Application Number Priority Date Filing Date Title
CN202180042708.1A CN115702288A (en) 2020-07-06 2021-07-06 Engine abnormality diagnosis method, engine abnormality diagnosis program, and engine abnormality diagnosis system
JP2022535361A JP7450238B2 (en) 2020-07-06 2021-07-06 Engine abnormality diagnosis method, engine abnormality diagnosis program, and engine abnormality diagnosis system
KR1020227043363A KR20230009484A (en) 2020-07-06 2021-07-06 Engine abnormality diagnosis method, engine abnormality diagnosis program, and engine abnormality diagnosis system

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2020116438 2020-07-06
JP2020-116438 2020-07-06

Publications (1)

Publication Number Publication Date
WO2022009904A1 true WO2022009904A1 (en) 2022-01-13

Family

ID=79553234

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2021/025526 WO2022009904A1 (en) 2020-07-06 2021-07-06 Engine abnormality diagnosis method, engine abnormality diagnosis program, and engine abnormality diagnosis system

Country Status (4)

Country Link
JP (1) JP7450238B2 (en)
KR (1) KR20230009484A (en)
CN (1) CN115702288A (en)
WO (1) WO2022009904A1 (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102645702B1 (en) * 2023-04-10 2024-03-11 사단법인 한국선급 Failure diagnosis system for marine engine combustion chamber amd method for diagnosing failure of marine engine combustion chamber using the same

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH10339664A (en) * 1997-06-10 1998-12-22 Babcock Hitachi Kk Monitor and monitoring method
WO2019175707A1 (en) * 2018-03-16 2019-09-19 株式会社半導体エネルギー研究所 Charge state estimation apparatus for secondary battery, abnormality detection apparatus for secondary battery, abnormality detection method for secondary battery, and management system for secondary battery

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5046104B2 (en) 2007-09-11 2012-10-10 独立行政法人 宇宙航空研究開発機構 Gas turbine engine performance estimation method and system
FR2978858B1 (en) 2011-08-01 2013-08-30 Airbus Operations Sas METHOD AND SYSTEM FOR DETERMINING FLIGHT PARAMETERS OF AN AIRCRAFT
US9436174B2 (en) 2013-03-01 2016-09-06 Fisher-Rosemount Systems, Inc. Kalman filters in process control systems

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH10339664A (en) * 1997-06-10 1998-12-22 Babcock Hitachi Kk Monitor and monitoring method
WO2019175707A1 (en) * 2018-03-16 2019-09-19 株式会社半導体エネルギー研究所 Charge state estimation apparatus for secondary battery, abnormality detection apparatus for secondary battery, abnormality detection method for secondary battery, and management system for secondary battery

Also Published As

Publication number Publication date
CN115702288A (en) 2023-02-14
JP7450238B2 (en) 2024-03-15
JPWO2022009904A1 (en) 2022-01-13
KR20230009484A (en) 2023-01-17

Similar Documents

Publication Publication Date Title
JP5575792B2 (en) Aircraft engine abnormality detection
US9465387B2 (en) Anomaly diagnosis system and anomaly diagnosis method
CN102246110B (en) Standardization of data used for monitoring an aircraft engine
EP1298512B1 (en) Adaptive aero-thermodynamic engine model
EP0858017B1 (en) Means and method for system performance tracking
US8417432B2 (en) Method for calculating confidence on prediction in fault diagnosis systems
WO2016133121A1 (en) Abnormality diagnosis method and abnormality diagnosis system
US20120310597A1 (en) Failure cause diagnosis system and method
WO2017209179A1 (en) Monitoring device, method for monitoring target device, and program
EP2509042A1 (en) Methods and systems for monitoring operation of equipment
CN113053171B (en) Civil aircraft system risk early warning method and system
EP1656597B1 (en) method and flight management system for mathematically modeling the performance characteristics of an aircraft
EP3191978A1 (en) Apparatus and method for ensembles of kernel regression models
WO2022009904A1 (en) Engine abnormality diagnosis method, engine abnormality diagnosis program, and engine abnormality diagnosis system
CN114048642B (en) Method for analyzing performance trend of aero-engine
EP2583146A1 (en) Asset health monitoring using stable distributions for heavy-tailed data
JP6404700B2 (en) Sensor combination determination device and sensor combination determination method
WO2023054039A1 (en) Performance evaluation method, performance evaluation program, and performance evaluation system through actual ship monitoring analysis of ship
DK180041B1 (en) Device, program, recording medium, and method for determining whether status of load bearing device is normal or abnormal
CN113821893A (en) Adaptive state estimation method for servo actuation system of aircraft engine
WO2005050337A2 (en) Intelligent database for performance predictions
US20230415921A1 (en) Method for determining an efficiency fault of an aircraft turboshaft engine module
JP2024064562A (en) Method for monitoring performance deterioration of ship propulsion system, program for monitoring performance deterioration, and system for monitoring performance deterioration
Schadler et al. Distance-Based Data Selection for Regression Models Used in an Adaptive Fault Detection Approach
Yannoulis et al. SOFTWARE TOOL FOR RAPID EVALUATION OF MARINE ENGINE PERFORMANCE IN SERVICE

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21838410

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022535361

Country of ref document: JP

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 20227043363

Country of ref document: KR

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21838410

Country of ref document: EP

Kind code of ref document: A1