CN110717630A - Reliability prediction method and device for turbine overspeed protection system - Google Patents

Reliability prediction method and device for turbine overspeed protection system Download PDF

Info

Publication number
CN110717630A
CN110717630A CN201910955332.8A CN201910955332A CN110717630A CN 110717630 A CN110717630 A CN 110717630A CN 201910955332 A CN201910955332 A CN 201910955332A CN 110717630 A CN110717630 A CN 110717630A
Authority
CN
China
Prior art keywords
characteristic parameter
fault
key device
reliability
membership
Prior art date
Legal status (The legal status 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 status listed.)
Granted
Application number
CN201910955332.8A
Other languages
Chinese (zh)
Other versions
CN110717630B (en
Inventor
吴志钢
魏振华
陈思沛
林令知
陈雯
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Nuclear Electric Power Planning Design and Research Institute Co Ltd
Original Assignee
State Nuclear Electric Power Planning Design and Research Institute Co Ltd
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 State Nuclear Electric Power Planning Design and Research Institute Co Ltd filed Critical State Nuclear Electric Power Planning Design and Research Institute Co Ltd
Priority to CN201910955332.8A priority Critical patent/CN110717630B/en
Publication of CN110717630A publication Critical patent/CN110717630A/en
Application granted granted Critical
Publication of CN110717630B publication Critical patent/CN110717630B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/20—Administration of product repair or maintenance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/04—Manufacturing
    • 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
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • General Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Operations Research (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Quality & Reliability (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Molecular Biology (AREA)
  • Game Theory and Decision Science (AREA)
  • Manufacturing & Machinery (AREA)
  • Development Economics (AREA)
  • Primary Health Care (AREA)
  • Testing And Monitoring For Control Systems (AREA)
  • Control Of Turbines (AREA)

Abstract

The invention discloses a reliability prediction method and a reliability prediction device for a turbine overspeed protection system, which relate to the technical field of reliability evaluation, and are characterized in that a regression neural network model for predicting the reliability of the turbine overspeed protection system is trained according to a plurality of first characteristic parameter sets in a period of time of the system; acquiring a second initial characteristic parameter set of the system at a first moment; and inputting the second initial characteristic parameter set into the trained recurrent neural network model to obtain the reliability of the system at the second moment. The invention discloses a reliability prediction method and a reliability prediction device for a turbine overspeed protection system, which are used for realizing dynamic prediction of the reliability of the turbine overspeed protection system in a multi-fault concurrent state.

Description

Reliability prediction method and device for turbine overspeed protection system
Technical Field
The invention relates to the technical field of reliability evaluation, in particular to a reliability prediction method and a reliability prediction device for a turbine overspeed protection system.
Background
The turbine overspeed protection system is an important guarantee for safe operation of a nuclear power unit, the reliable action of the turbine overspeed protection system is supported by the effective action of each device in the system, and the system reliability is reduced and even the system cannot act due to the fact that any device fails. And along with long-term operation of the steam turbine set, various faults of different degrees may occur to each device, and influence of different degrees is generated on the system. Through the prediction of the reliability of the turbine overspeed protection system in various fault states, the safe and stable operation of the unit can be guaranteed, reliable suggestions can be provided for system maintenance, and the economic benefits of system operation are improved.
In the related technology, the reliability research of the overspeed protection system of the nuclear turbine is mainly based on theoretical analysis and engineering research, so that reasonable suggestions are provided for maintenance and improvement of the system.
In the process of implementing the invention, the inventor finds that the related art has at least the following problems: the traditional reliability analysis method cannot dynamically predict the reliability of the overspeed protection system of the nuclear turbine.
Disclosure of Invention
The embodiment of the invention provides a reliability prediction method and a reliability prediction device for a turbine overspeed protection system, which can realize the reliability prediction of the turbine overspeed protection system in a multi-fault concurrent state. The specific technical scheme is as follows:
a reliability prediction method for a turbine overspeed protection system comprises the following steps:
training a regression neural network model for predicting the reliability of the turbine overspeed protection system according to a plurality of first characteristic parameter sets in the system within a period of time;
acquiring a second initial characteristic parameter set of the system at a first moment;
and inputting the second initial characteristic parameter set to the trained recurrent neural network model to obtain the reliability of the system at the second moment.
In one implementation manner of the embodiment of the present application, the training of the recurrent neural network model for predicting the reliability of the turbine overspeed protection system according to the plurality of first feature parameter sets in the system over a period of time includes:
acquiring a first characteristic parameter set corresponding to different sampling moments in a period of time of the system and the reliability of the system corresponding to the first characteristic parameter set;
and inputting the plurality of first characteristic parameter sets into the recurrent neural network model according to the corresponding sampling sequence, and training the recurrent neural network model by taking the reliability of the corresponding system as target output.
In an implementation manner of the embodiment of the present application, the determining a first feature parameter set corresponding to different sampling moments in a period of time of the system includes:
determining key equipment with faults at the current sampling moment of the system;
acquiring a plurality of first initial characteristic parameter sets corresponding to the failed key equipment;
determining fuzzy weight matrixes corresponding to the plurality of first initial characteristic parameter sets according to the failed key equipment;
and determining a first characteristic parameter set corresponding to the current sampling moment according to the fuzzy weight matrix and the plurality of first initial characteristic parameter sets.
In an implementation manner of the embodiment of the present application, the determining, according to the failed key device, the fuzzy weight matrix corresponding to the plurality of first initial feature parameter sets includes:
acquiring the corresponding relation between key equipment and the fuzzy weight matrix;
and determining fuzzy weight matrixes corresponding to the plurality of first initial characteristic parameter sets according to the corresponding relation.
In an implementation manner of the embodiment of the present application, before obtaining a corresponding relationship between a key device and the fuzzy weight matrix, the method further includes:
determining a fuzzy weight matrix corresponding to each key device;
and pre-establishing a corresponding relation between the key equipment and the fuzzy weight matrix.
In an implementation manner of the embodiment of the present application, the determining a fuzzy weight matrix corresponding to each of the key devices includes:
determining a fault category corresponding to each first initial characteristic parameter set in a plurality of first initial characteristic parameter sets corresponding to each key device;
obtaining a first membership matrix of each fault category of each key device according to a first initial characteristic parameter set corresponding to each fault category of each key device;
normalizing the first membership degree corresponding to each characteristic parameter in the first membership degree matrix corresponding to all fault types of each key device to obtain the weight corresponding to each characteristic parameter of each fault type of each key device;
and combining the weights corresponding to the characteristic parameters of all fault types of each key device to obtain a fuzzy weight matrix corresponding to each key device.
In an implementation manner of the embodiment of the present application, the obtaining a first membership matrix of each fault category of each key device according to a first initial feature parameter set corresponding to each fault category of each key device includes the following steps:
determining the fault degree corresponding to each first initial characteristic parameter set in a plurality of first initial characteristic parameters corresponding to each fault category of each key device;
obtaining a second membership matrix of each fault degree of each fault category of each key device according to the first initial characteristic parameter set corresponding to each fault degree of each fault category of each key device;
averaging the second membership degrees corresponding to each characteristic parameter in the second membership degree matrix corresponding to each fault degree of each fault category of each key device to obtain a first membership degree corresponding to the characteristic parameter;
and combining the first membership corresponding to each characteristic parameter to obtain a first membership matrix corresponding to each fault type of each key device.
In an implementation manner of the embodiment of the present application, obtaining a second membership matrix of each fault degree of each fault category of each key device according to a first initial feature parameter set corresponding to each fault degree of each fault category of each key device includes:
determining a second membership degree corresponding to each characteristic parameter by using a normal function as a fuzzy operator;
and combining the second membership corresponding to each characteristic parameter to obtain the second membership matrix.
In an implementation manner of the embodiment of the present application, determining the second membership degree corresponding to each of the characteristic parameters by using a normal function as a fuzzy operator includes:
counting each characteristic parameter in a plurality of characteristic parameter sets corresponding to different fault degrees of the same key equipment and the same fault type, and determining a normal function corresponding to each characteristic parameter;
and calculating a second membership degree corresponding to each characteristic parameter according to the value of each characteristic parameter.
The utility model provides a reliability prediction unit of steam turbine overspeed protection system, includes:
a training module configured to train a recurrent neural network model for predicting reliability of a turbine overspeed protection system according to a plurality of first feature parameter sets of the system over a period of time;
an obtaining module configured to obtain a second initial feature parameter set of the system at a first time;
and the prediction module is configured to input the second initial characteristic parameter set to the trained recurrent neural network model to obtain the reliability of the system at a second moment.
The beneficial effects of the embodiment of the application at least comprise:
according to the reliability prediction method and the reliability prediction device for the turbine overspeed protection system, the regression neural network model is trained according to a plurality of first characteristic parameter sets of the system, and the reliability of the system at the next moment is calculated according to the trained regression neural network model, so that the dynamic prediction of the reliability of the turbine overspeed protection system in a multi-fault concurrent state is realized.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a schematic flow chart illustrating a method for predicting the reliability of an overspeed protection system of a steam turbine according to an embodiment of the present invention;
fig. 2 is a schematic flowchart of step S101 in fig. 1 according to an embodiment of the present invention;
fig. 3 is a schematic flowchart of step S201 in fig. 2 according to an embodiment of the present invention;
fig. 4 is a schematic flowchart of a process of determining a fuzzy weight matrix corresponding to each key device according to an embodiment of the present invention;
fig. 5 is a schematic flowchart of step S402 in fig. 4 according to an embodiment of the present invention;
fig. 6 is a flowchart illustrating step S502 in fig. 5 according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
The turbine is a high-speed rotating device, the centrifugal force of a rotating part is in direct proportion to the square of the rotating speed, when the rotating speed of the turbine exceeds 20% of the rated rotating speed, the centrifugal stress is close to 1.5 times of the stress under the rated rotating speed, at the moment, the rotating part loosens, and simultaneously, the centrifugal force exceeds the strength limit allowed by materials to damage the part, therefore, the turbine is provided with an overspeed protection device which can act when the rotating speed of the turbine exceeds 10-12% of the rated rotating speed, and the air inlet of the turbine is cut off and the turbine stops working rapidly.
The turbine overspeed protection can be divided into three layers: the first layer is an Overspeed Protection Controller (OPC), and when the action rotating speed reaches 103% of the rated rotating speed, the OPC electromagnetic valve acts; the second layer is Automatic Stop Trip (AST) protection, and when the action rotating speed reaches 110% of the rated rotating speed, the AST electromagnetic valve acts.
When the turbine overspeed protection system operates normally, the OPC electromagnetic valve is not electrified and is normally closed, and a discharge channel of the OPC main pipe is closed, so that the lower cavity of a piston of an actuating mechanism of the high-pressure and medium-pressure regulating steam valve can establish oil pressure. Once the OPC operation condition is reached, the OPC electromagnetic valve is electrified and opened, so that the oil in the OPC main pipe is discharged, and thus, the unloading valve on the actuating mechanism of the high-pressure and medium-pressure regulating steam valve is quickly opened, and the high-pressure and medium-pressure regulating steam valve is quickly closed.
When the AST electromagnetic valve normally operates, the AST electromagnetic valve is in a power-on closing state, so that an oil drainage channel on an AST main pipe is closed, and the oil pressure of piston lower cavities of all steam valve actuating mechanisms can be established. When the rotating speed of the steam turbine reaches 110% of the rated rotating speed, the AST electromagnetic valve acts to close the high-pressure main steam valve and the medium-pressure main steam valve, meanwhile, a check valve connected with the OPC oil loop is jacked open to enable the OPC oil circuit to release pressure, and the high-pressure regulating steam valve and the medium-pressure regulating steam valve are closed to realize emergency stop.
The equipment in the turbine overspeed protection system is complex in composition, and various faults of different degrees can occur to each equipment, so that various fault states occur, and influence of different degrees is generated on the reliability of the system.
The embodiment of the application provides a reliability prediction method of a turbine overspeed protection system, which is executed by a computer. As shown in fig. 1, the method comprises the steps of:
s101, training a regression neural network model for predicting the reliability of the turbine overspeed protection system according to a plurality of first characteristic parameter sets in a period of time;
s102, acquiring a second initial characteristic parameter set of the system at a first moment;
and S103, inputting the second initial characteristic parameter set to the trained recurrent neural network model to obtain the reliability of the system at the second moment.
In the embodiment of the present application, the second time is a time after the first time. The recurrent neural network is a dynamic network formed by static neurons and network output feedback, and an accurate estimation model can be established by combining the advantages of the recurrent neural network and the neural network. In the embodiment of the present application, a nonlinear autoregressive model (NARX) with external input may be adopted, and the model includes an input layer, an implicit layer, an output layer, an input delay, and an output delay.
The delay of the NARX recurrent neural network model affects the prediction effect of the recurrent neural network model, and different input delays determine the input data for prediction. For example, when the sample time interval of the training samples is 1 day when the NARX regression neural network model is trained, and the input delay is set to 1, the second initial feature parameter set acquired yesterday is required to be input to predict the reliability of the system today. When the input delay is set to 2, the second initial feature parameter set acquired two days before the input is needed to predict the reliability of the system of today.
According to the reliability prediction method for the turbine overspeed protection system, the regression neural network model is trained according to a plurality of first characteristic parameter sets of the system, and the reliability of the system at the next moment is calculated according to the trained regression neural network model, so that the dynamic prediction of the reliability of the turbine overspeed protection system in a multi-fault concurrent state is realized.
In the embodiment of the present application, as shown in fig. 2, the training of the recurrent neural network model for predicting the reliability of the turbine overspeed protection system according to the plurality of first feature parameter sets in the system for a period of time in step S101 may include the following steps:
s201, acquiring a first feature parameter set corresponding to different sampling moments in a period of time of a system and the reliability of the corresponding system;
in this step, the first feature parameter set is statistics of different fault types corresponding to the key device that has a fault at the current sampling time, so that the first feature parameter set can better represent the multi-fault concurrence state of the system. The regression neural network model is trained according to the first characteristic parameter set and the reliability of the system corresponding to the first characteristic parameter set, so that the reliability of the turbine overspeed protection system in a multi-fault concurrent state can be more accurately predicted.
In the embodiment of the present application, the reliability of the system corresponding to each of the plurality of first feature parameter sets is obtained, and first, a fault state of the system corresponding to each of the first feature parameter sets needs to be determined, and then, the reliability of the system in the fault state is calculated. The fault state of the system comprises a critical device with a fault, a fault type and a fault degree. The fault state of the system is obtained according to the first characteristic parameter set and can be determined by historical fault record data of the overspeed protection system of the steam turbine of the nuclear power plant, the first characteristic parameter set is compared with the historical fault record data, and the fault state corresponding to data which is close to the first characteristic parameter set in the historical fault record data is determined as the fault state corresponding to the first characteristic parameter set. In this step, the reliability of the system in a certain fault state may be calculated in various ways.
In some embodiments of the present application, a fault tree method is used to perform the reliability calculation of the system. The method uses an unexpected event as a vertex, and finds out all direct reasons and indirect reasons of the event one by one through layer analysis from top to bottom, establishes a logical relation and draws a tree-like graph. And determining the occurrence probability of each bottom layer event, and solving the occurrence probability of the top event according to the fault tree.
Illustratively, the turbine overspeed protection system is completely failed, and a destructive overspeed accident occurs as a top event, and the direct reason for the occurrence of the event is that the main oil pump fails to supply oil normally. Possible reasons for the main oil pump failing to supply oil normally include power loss, runaway, abrasion of the main oil pump and blockage of a filter screen on an oil way. The occurrence probability of each bottom layer event can be determined according to fault statistical data of the overspeed protection system of the steam turbine of the nuclear power plant or by combining data recorded in published literature in the field.
In other embodiments of the present application, a GO method is used to calculate the reliability of the system. The reliability of the system in various fault states is calculated by adopting a GO method, and the method comprises the following steps:
establishing a GO model of the system, and converting the GO model into a GO graph;
and respectively calculating the reliability of the system when no fault, single fault and multiple faults occur based on the logical relationship among the devices in the GO diagram and by combining the reliability parameters of the devices.
In the embodiment of the application, before the GO model of the system is established, the key equipment serving as a research object in the system needs to be determined, and the GO model of the system is established based on the selected key equipment.
In the embodiment of the application, the selected key equipment of the system comprises a main oil pump, a standby oil pump, a filter screen, an OPC electromagnetic valve, an AST electromagnetic valve and a quick unloading valve. When the GO map of the system is established, an oil supply subsystem, a security subsystem and an execution subsystem in the system are used as frameworks, and effective actions of the system are used as successful guidance.
Specifically, in this embodiment, the main oil pump, the backup oil pump, and the filter screen on the oil path are used as oil supply subsystems to supply oil to other devices. There is the stand-by relation between main oil pump and the reserve oil pump, and when the main oil pump broke down, the reserve oil pump just started, consequently, in the GO picture of this embodiment, main oil pump and reserve oil pump were located two parallelly connected branches respectively, still establish ties the filter screen on every branch. The main oil pump and the backup oil pump are electric pumps, and electric signals are input.
The oil liquid of the main oil pump and the standby oil pump is supplied to the OPC electromagnetic valve and the AST electromagnetic valve through a filter screen, and the OPC electromagnetic valve and the AST electromagnetic valve are used as security subsystems to control the action of the quick unloading valve. The quick unloading valve is used as an execution subsystem and can realize the opening and closing of the corresponding steam valve. The OPC electromagnetic valve is electrified and opened, so that oil in the OPC main pipe is discharged, the quick unloading valve on the regulating steam valve is opened, and the regulating steam valve is closed. When the rotating speed of the steam turbine exceeds the adjustable range of the OPC electromagnetic valve, the AST electromagnetic valve is opened, the quick unloading valve on the main steam valve is opened, and the main steam valve is closed. Meanwhile, a check valve connected with the OPC oil loop acts, an OPC oil circuit releases pressure, a high-pressure regulating steam valve and a medium-pressure regulating steam valve are closed, and emergency shutdown is realized. It will be appreciated that multiple OPC solenoid valves and multiple AST solenoid valves may be included in the security subsystem, and only one OPC solenoid valve and one AST solenoid valve will be described here as representative.
In the GO diagram of this embodiment, the OPC solenoid valve and the quick unloading valve are connected in series on the same branch, the AST solenoid valve, the unloading valve of the main steam valve, the OPC solenoid valve, and the unloading valve of the adjustment steam valve are connected in series in sequence on the other branch, and the two branches are connected in parallel. When the two branches in the GO diagram are all conducted, the system can effectively act, namely, the turbine overspeed protection system can normally operate. And calculating the reliability of the system in the fault state according to the probability of the fault of each key device in the GO diagram. Each acquired first initial characteristic parameter set corresponds to a determined fault state of the system, and the first initial characteristic parameter set, the second initial characteristic parameter set and the system are in a mutual corresponding relation according to the reliability of the system corresponding to the fault state. The probability of a certain fault can be determined according to fault statistical data of the overspeed protection system of the steam turbine of the nuclear power plant or by combining data recorded in published documents in the field.
In the embodiment of the application, the reliability of the system in a fault-free state can be determined according to the GO map, namely the basic reliability, so that the reliability of the system in a certain fault state is compared with the reliability in the fault-free state, and a technician can be prompted to repair when the deviation between the reliability in the fault state and the basic reliability is large. Meanwhile, when the deviation between the predicted reliability of the system at the next moment and the basic reliability is large, a technician can repair the system in advance.
S202, inputting the plurality of first characteristic parameter sets into the recurrent neural network model according to the corresponding sampling sequence, and training the recurrent neural network model by taking the reliability of the corresponding system as target output.
In the embodiment of the present application, a plurality of first initial feature parameter sets are obtained at fixed time intervals. Illustratively, the system is sampled at the same time of day during a period of time, i.e. the sampling interval between two adjacent first feature parameter sets is one day.
Further, the obtained plurality of first feature parameter sets are divided into training samples, verification samples and test samples to train the recurrent neural network model. And inputting the first characteristic parameter set into the recurrent neural network model according to the corresponding sampling sequence, and training the recurrent neural network model by taking the first characteristic parameter set as an input sample and the reliability of the system corresponding to the first characteristic parameter set as target output.
In this embodiment of the application, as shown in fig. 3, the determining, in step S201, a first feature parameter set corresponding to different sampling times within a period of time of the system may include the following steps:
s301, determining key equipment with faults at the current sampling moment of the system;
in the step, a turbine overspeed protection system can be selected as a sampling standard, a group of standard characteristic parameter sets of the system at the current sampling moment are obtained, and the key equipment with faults at the current sampling moment is determined according to the standard characteristic parameter sets. In some embodiments of the present application, a monitoring system may be provided in the system, and the monitoring system may directly obtain the operation states of the devices in the turbine overspeed protection system, and determine the critical device that has a fault according to the device with an abnormal operation state. In other embodiments of the present application, the historical fault record data or the simulation model of the overspeed protection system of the nuclear power plant turbine can also be used for determining the historical fault record data or the simulation model. The key equipment with failure can comprise a main oil pump, a standby oil pump, a filter screen, an OPC electromagnetic valve, an AST electromagnetic valve and a quick unloading valve.
S302, acquiring a plurality of first initial characteristic parameter sets corresponding to the key equipment with the fault;
in this step, the plurality of first initial feature parameter sets corresponding to the key device may characterize all possible fault states of the key device at the current sampling time, where the fault state of each key device includes a fault category and a fault degree. In some embodiments of the present application, a plurality of first initial characteristic parameter sets of different fault states of the critical equipment may be obtained by sampling a plurality of different turbine overspeed protection systems under the same operating conditions. In other embodiments of the present application, the plurality of first initial characteristic parameter sets corresponding to the failed critical device may also be obtained from historical fault record data of the overspeed protection system of the steam turbine of the nuclear power plant.
The first initial set of feature parameters may include a plurality of feature parameters for predicting system reliability. Based on the structure of the turbine overspeed protection system, the first initial characteristic parameter set in the embodiment of the present application may include a main oil path pressure difference, a backup oil path pressure difference, a high-pressure oil line oil pressure, an oil return line oil pressure, and an emergency trip oil pressure. Wherein, the pressure difference of the main oil circuit refers to the pressure difference of the oil pressure in front of and behind the filter screen on the main oil circuit. The pressure difference of the spare oil way refers to the pressure difference of oil pressure in front of and behind the filter screen on the spare oil way. The oil pressure of the high-pressure oil pipeline refers to the oil pressure of the high-pressure oil pipeline in the system. The oil return line oil pressure is the oil pressure on the oil return line in the system. The critical shutoff oil pressure is the oil pressure established when the AST solenoid valve is closed. Each characteristic parameter in the first initial characteristic parameter set can be directly displayed according to a monitoring system connected with the system, or can be acquired through a testing device arranged at each testing point on the system.
S303, determining fuzzy weight matrixes corresponding to the plurality of first initial characteristic parameter sets according to the key equipment with the fault;
in this step, after acquiring a plurality of first initial feature parameter sets corresponding to the failed key device, the fuzzy weight matrix corresponding to each key device may be determined, so as to pre-establish a corresponding relationship between the key device and the fuzzy weight matrix. When determining the fuzzy weight matrix corresponding to the first initial feature parameter set, the corresponding relationship may be obtained, and fuzzy weight matrices corresponding to the plurality of first initial feature parameter sets may be determined according to the corresponding relationship.
Further, a flag corresponding to each critical device may be pre-established, and after determining the failed critical device corresponding to each first initial characteristic parameter, the flag of the corresponding critical device is attached to the first initial characteristic parameter set. After the corresponding relation between the key equipment and the fuzzy weight matrix is established, the corresponding relation is stored, so that the corresponding fuzzy weight matrix can be directly obtained according to the mark attached to the first initial characteristic parameter set.
S304, determining a first characteristic parameter set corresponding to the current sampling moment according to the fuzzy weight matrix and the plurality of first initial characteristic parameter sets.
In this step, the fuzzy weight matrix corresponding to the failed key device includes the weight of each feature parameter under different failure types of the key device. The plurality of first initial characteristic parameter sets respectively correspond to different fault types of the key equipment, so that the weighted sum of each characteristic parameter in the plurality of characteristic parameter sets and the corresponding weight value is the characteristic parameter value in the first characteristic parameter set.
In the embodiment of the present application, as shown in fig. 4, determining the fuzzy weight matrix corresponding to each key device includes the following steps:
s401, determining a fault type corresponding to each first initial characteristic parameter set in a plurality of first initial characteristic parameter sets corresponding to each key device;
s402, obtaining a first membership matrix of each fault category of each key device according to a first initial characteristic parameter set corresponding to each fault category of each key device;
s403, normalizing the first membership degree corresponding to each characteristic parameter in the first membership degree matrix corresponding to all fault types of each key device to obtain a weight corresponding to each characteristic parameter of each fault type of each key device;
s404, combining the weights corresponding to the characteristic parameters of all fault types of each key device to obtain a fuzzy weight matrix corresponding to each key device.
In the embodiment of the application, the types of possible faults of the main oil pump and the backup oil pump can be the same, including power loss, runaway and abrasion. The possible fault type of the filter screen is blockage. The types of faults that may occur with OPC solenoid valves and AST solenoid valves include loss of power, loss of control, wear, and oil leakage. The types of faults that can occur in a quick unloading valve include runaway, wear and oil leakage. The fault type corresponding to each first initial characteristic parameter set can be determined according to the simulation model and historical fault record data of the overspeed protection system of the steam turbine of the nuclear power plant and the like.
Further, because different fault types may exist when each key device fails, and the occurrence probability of each fault type is different, the first membership degree of each characteristic parameter corresponding to each fault type may be obtained first, and the first membership degrees of the characteristic parameters corresponding to different fault types are normalized, so as to obtain the weight corresponding to the characteristic parameter when the fault type occurs.
In this embodiment of the application, as shown in fig. 5, the obtaining, in step S402, a first membership matrix of each fault category of each key device according to a first initial feature parameter set corresponding to each fault category of each key device includes the following steps:
s501, determining the fault degree corresponding to each first initial characteristic parameter set in a plurality of first initial characteristic parameters corresponding to each fault type of each key device;
s502, obtaining a second membership degree matrix of each fault degree of each fault type of each key device according to the first initial characteristic parameter set corresponding to each fault degree of each fault type of each key device;
s503, averaging the second membership degrees corresponding to each characteristic parameter in the second membership degree matrix corresponding to each fault degree of each fault type of each key device to obtain a first membership degree corresponding to the characteristic parameter;
s504, the first membership degrees corresponding to the characteristic parameters are combined to obtain a first membership degree matrix corresponding to each fault type of each key device.
In the embodiment of the application, the same key equipment and the same fault category can be divided into multiple fault degrees, and the second membership degree matrix corresponding to different fault degrees can represent the probability of the fault degree occurring in the fault category to which the fault degree belongs. And averaging the second membership degrees of the characteristic parameters corresponding to different fault degrees, so as to obtain the first membership degree of the characteristic parameter corresponding to each fault type of each key device.
Each fault type can be divided into four fault degrees of emergency, serious, general and slight, and the fault type can be expressed by taking values within the interval of 1-100 from no fault to complete fault. The larger the value is, the more serious the fault degree is, and the complete fault means that the key equipment completely fails. The degree of failure corresponding to each first set of initial feature parameters may be artificially partitioned. Further, the degree of failure may also be attached as a degree mark to the corresponding first initial characteristic parameter set.
In the embodiment of the present application, as shown in fig. 6, the determining, in step S502, a second membership matrix corresponding to the same fault degree of the same key device and the same fault type includes the following steps:
s601, determining a second membership degree corresponding to each characteristic parameter by using a normal function as a fuzzy operator;
and S602, combining the second membership corresponding to each characteristic parameter to obtain a second membership matrix.
In this step, a normal function is used as a fuzzy operator, each characteristic parameter in a plurality of characteristic parameter sets corresponding to different fault degrees of the same key device and the same fault type can be counted, and the normal function corresponding to each characteristic parameter is determined according to a distribution graph of the characteristic parameter. And after the normal function corresponding to the characteristic parameter is determined, taking the value of the characteristic parameter corresponding to each fault degree as the independent variable to be brought into the normal function corresponding to the characteristic parameter, wherein the function value of the normal function is the second membership degree corresponding to the characteristic parameter.
When each characteristic parameter in a plurality of characteristic parameter sets corresponding to different fault degrees of the same key equipment and the same fault type is counted, a distribution graph is drawn by taking the characteristic parameter value as an abscissa and the fault degree corresponding to the characteristic parameter value as an ordinate. And determining a normal function corresponding to the characteristic parameter according to the characteristics of the distribution graph, such as the symmetry axis, the steepness degree and the like.
In another embodiment of the present application, a method for predicting reliability of an overspeed protection system of a steam turbine is provided, comprising the steps of:
s701, determining key equipment with faults at the current sampling moment of the system;
s702, acquiring a plurality of first initial characteristic parameter sets corresponding to the key equipment with the fault;
s703, determining a fault type corresponding to each first initial characteristic parameter set in a plurality of first initial characteristic parameter sets corresponding to each key device;
s704, determining the fault degree corresponding to each first initial characteristic parameter set in a plurality of first initial characteristic parameters corresponding to each fault type of each key device;
there are multiple key devices in the system, and the processing modes of the devices are the same, and the following takes the example of obtaining the fuzzy weight matrix of the key device a.
Assuming that n common fault categories exist in the key equipment A, the set of the fault categories of the equipment A is
Figure BDA0002227088800000121
Assuming that each fault category is divided into d fault degrees, the set of fault degrees is
Figure BDA0002227088800000122
Suppose that in the j-th failure case of the A device, a certain first initial characteristic parameter set is represented as
Figure BDA0002227088800000123
And is
Figure BDA0002227088800000124
Wherein i represents a characteristic parameter number, j represents a fault type number, and t represents a fault degree number.
S705, determining a second membership degree corresponding to each characteristic parameter by using a normal function as a fuzzy operator;
using normal functions
Figure BDA0002227088800000125
And as a fuzzy operator, determining parameters alpha and sigma in the normal function according to the characteristics of the distribution graph of each characteristic parameter.
A device, j fault type and DtIth characteristic parameter corresponding to fault degree
Figure BDA0002227088800000126
Is substituted into the normal function corresponding to the characteristic parameter, and the independent variable value of the normal function is taken as the characteristic parameter
Figure BDA0002227088800000131
Second degree of membership of
Figure BDA0002227088800000132
S706, combining the second membership degrees corresponding to each characteristic parameter to obtain a second membership degree matrix;
obtaining a second membership degree of each characteristic parameter in the first initial characteristic parameter set, and obtaining a second membership degree matrix corresponding to the failure type of the equipment A
Figure BDA0002227088800000133
And is
Figure BDA0002227088800000134
Wherein i represents a characteristic parameter number, j represents a fault type number, and t represents a fault degree number.
S707, averaging the second membership degrees corresponding to each characteristic parameter in the second membership degree matrix corresponding to each fault degree of each fault type of each key device to obtain a first membership degree corresponding to the characteristic parameter;
equipment A, fault type j, and DtThe second corresponding to the fault degreei characteristic parameters
Figure BDA0002227088800000135
Second degree of membership ofTaking an average value to obtain
Figure BDA0002227088800000137
Then
Figure BDA0002227088800000138
And the first membership degree of the ith characteristic parameter corresponding to the equipment A and the j fault type.
S708, combining the first membership degrees corresponding to the characteristic parameters to obtain a first membership degree matrix corresponding to each fault type of each key device;
the first membership matrix corresponding to the equipment A and the fault category j is as follows:
Figure BDA0002227088800000139
wherein i represents a characteristic parameter number, and j represents a fault type number.
S709, normalizing the first membership degree corresponding to each characteristic parameter in the first membership degree matrix corresponding to all fault types of each key device to obtain a weight corresponding to each characteristic parameter of each fault type of each key device;
combining a plurality of first membership degree matrixes corresponding to different fault types of the equipment A, wherein the membership degree matrix of the equipment A is as follows:
Figure BDA00022270888000001310
wherein i represents a characteristic parameter number, and j represents a fault type number.
Corresponding the first membership degree of the ith characteristic parameter of the equipment A and the fault type j
Figure BDA0002227088800000141
Normalized to obtainThen c isiAnd the weights of the ith characteristic parameters corresponding to the equipment A and the j fault types are obtained.
S710, combining weights corresponding to the characteristic parameters of all fault types of each key device to obtain a fuzzy weight matrix corresponding to each key device;
combining the weight of each characteristic parameter to obtain a fuzzy weight matrix of the equipment A:
Figure BDA0002227088800000143
and is
Figure BDA0002227088800000144
Wherein i represents a characteristic parameter number, and j represents a fault type number.
S711, determining a first feature parameter set corresponding to the current sampling moment according to the fuzzy weight matrix and the plurality of first initial feature parameter sets;
and on the basis of obtaining the fuzzy weight matrix of the equipment A, carrying out fuzzy weighted summation on the first initial characteristic parameter sets corresponding to different fault types of the equipment A to obtain the first characteristic parameter set corresponding to the equipment A. The ith characteristic parameter in the first characteristic parameter set corresponding to the current sampling time is:
Figure BDA0002227088800000145
wherein i represents a characteristic parameter number, and j represents a fault type number.
S712, obtaining the reliability of the system corresponding to each of the plurality of first feature parameter sets;
s713, inputting the plurality of first characteristic parameter sets into the recurrent neural network model according to the corresponding sampling sequence, and training the recurrent neural network model by taking the reliability of the corresponding system as target output;
s714, acquiring a second initial characteristic parameter set of the system at the first moment;
and S715, inputting the second initial characteristic parameter set to the trained recurrent neural network model to obtain the reliability of the system at the second moment.
According to the method and the device, a first initial characteristic parameter set obtained by key equipment in a multi-fault concurrent state is subjected to weighted correction by using a fuzzy weight matrix of the key equipment to obtain the first characteristic parameter set, a regression neural network model is trained by using the first initial characteristic parameter set and the reliability of a corresponding system, and the reliability of the system at the next moment is predicted by using the regression neural network model, so that the dynamic prediction of the reliability of the turbine overspeed protection system in the multi-fault concurrent state is realized.
Based on the same technical concept, the embodiment of the present application further provides a reliability prediction device for a turbine overspeed protection system, including:
the training module is used for training a regression neural network model for predicting the reliability of the turbine overspeed protection system according to a plurality of first characteristic parameter sets in a period of time;
the acquisition module is configured to acquire a second initial characteristic parameter set of the system at a first moment;
and the prediction module is configured to input the second initial characteristic parameter set to the trained recurrent neural network model to obtain the reliability of the system at the second moment.
In an embodiment of the present application, the training module includes:
the first obtaining submodule is configured to obtain a first feature parameter set corresponding to different sampling moments in a period of time of the system and the reliability of the corresponding system;
and the training sub-module is configured to input the plurality of first characteristic parameter sets into the recurrent neural network model according to the corresponding sampling sequence of the plurality of first characteristic parameter sets, and train the recurrent neural network model by taking the reliability of the corresponding system as target output.
In this embodiment of the present application, the first obtaining sub-module includes:
the first determining submodule is configured to determine key equipment with faults at the current sampling moment of the system;
the second obtaining sub-module is configured to obtain a plurality of first initial characteristic parameter sets corresponding to the failed key equipment;
the third obtaining submodule is configured to determine fuzzy weight matrixes corresponding to the plurality of first initial characteristic parameter sets according to the failed key equipment;
and the weighted summation sub-module is configured to determine a first feature parameter set corresponding to the current sampling moment according to the fuzzy weight matrix and the plurality of first initial feature parameter sets.
In this embodiment of the application, the third obtaining sub-module includes:
the fourth obtaining submodule is configured to obtain a corresponding relation between the key equipment and the fuzzy weight matrix;
and the second determining submodule is configured to determine fuzzy weight matrixes corresponding to the plurality of first initial characteristic parameter sets according to the corresponding relation.
In the embodiment of the present application, the apparatus further includes a corresponding relationship module, including:
the third determining submodule is configured to determine a fuzzy weight matrix corresponding to each key device;
and the relation establishing submodule is configured to establish a corresponding relation between the key equipment and the fuzzy weight matrix in advance.
In this embodiment of the present application, the third determining sub-module includes:
the fourth determining submodule is configured to determine a fault category corresponding to each first initial characteristic parameter set in a plurality of first initial characteristic parameter sets corresponding to each key device;
the first calculation submodule is configured to obtain a first membership matrix of each fault category of each key device according to a first initial characteristic parameter set corresponding to each fault category of each key device;
the normalizing calculation module is configured to normalize the first membership degree corresponding to each characteristic parameter in the first membership degree matrix corresponding to all fault types of each key device to obtain a weight corresponding to each characteristic parameter of each fault type of each key device;
and the first merging submodule is configured to merge the weights corresponding to the characteristic parameters of all the fault types of each key device to obtain a fuzzy weight matrix corresponding to each key device.
In an embodiment of the present application, the first calculation submodule includes:
the fifth determining submodule is configured to determine a fault degree corresponding to each first initial characteristic parameter set in the plurality of first initial characteristic parameters corresponding to each fault type of each key device;
the second calculation submodule is configured to obtain a second membership matrix of each fault degree of each fault category of each key device according to the first initial characteristic parameter set corresponding to each fault degree of each fault category of each key device;
the mean value submodule is configured to mean the second membership degrees corresponding to each characteristic parameter in the second membership degree matrix corresponding to each fault degree of each fault category of each key device to obtain a first membership degree corresponding to the characteristic parameter;
and the second merging submodule is configured to merge the first membership corresponding to each characteristic parameter to obtain a corresponding first membership matrix of each fault category of each key device.
In this embodiment of the present application, the second calculation submodule includes:
the sixth determining submodule is configured to determine a second membership degree corresponding to each characteristic parameter by using the normal function as a fuzzy operator;
and the third merging submodule is configured to merge the second membership corresponding to each characteristic parameter to obtain a second membership matrix.
In this embodiment of the present application, the sixth determining sub-module includes:
and the statistical distribution submodule is configured to count each characteristic parameter in a plurality of characteristic parameter sets corresponding to different fault degrees of the same key equipment and the same fault type, and determine a normal function corresponding to each characteristic parameter.
And the third calculation submodule is configured to calculate a second membership degree corresponding to each characteristic parameter according to the value of the characteristic parameter.
It should be noted that: when the reliability prediction device of the turbine overspeed protection system provided by the above embodiment performs reliability prediction, only the division of the above function modules is taken as an example, and in practical application, the function distribution can be completed by different function modules according to needs, that is, the internal structure of the device is divided into different function modules, so as to complete all or part of the above described functions. In addition, the reliability prediction apparatus and the reliability prediction method provided by the above embodiments belong to the same concept, and specific implementation processes thereof are described in the method embodiments and are not described herein again.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, and the program may be stored in a computer-readable storage medium, which may be a read-only memory, a magnetic disk or an optical disk, etc. For example, the reliability prediction method of the turbine overspeed protection system according to the present invention may be executed by a computer device, and the reliability prediction apparatus of the turbine overspeed protection system according to the present invention may also be a computer device. The computer device comprises a processor and a storage medium comprising program instructions which, when executed by the processor, implement the method of the above-described embodiments.
The above description is only exemplary of the invention and should not be taken as limiting the scope of the invention, which is intended to cover any variations, equivalents, or improvements included within the spirit and scope of the invention.

Claims (10)

1. A reliability prediction method for a turbine overspeed protection system is characterized by comprising the following steps:
training a regression neural network model for predicting the reliability of the turbine overspeed protection system according to a plurality of first characteristic parameter sets in the system within a period of time;
acquiring a second initial characteristic parameter set of the system at a first moment;
and inputting the second initial characteristic parameter set to the trained recurrent neural network model to obtain the reliability of the system at the second moment.
2. The method of claim 1, wherein training a regression neural network model for predicting reliability of a turbine overspeed protection system based on a plurality of first sets of signature parameters over a period of time in the system comprises:
acquiring a first characteristic parameter set corresponding to different sampling moments in a period of time of the system and the reliability of the system corresponding to the first characteristic parameter set;
and inputting the plurality of first characteristic parameter sets into the recurrent neural network model according to the corresponding sampling sequence, and training the recurrent neural network model by taking the reliability of the corresponding system as target output.
3. The method of claim 2, wherein the obtaining a first set of feature parameters corresponding to different sampling instants within a period of the system comprises:
determining key equipment with faults at the current sampling moment of the system;
acquiring a plurality of first initial characteristic parameter sets corresponding to the failed key equipment;
determining fuzzy weight matrixes corresponding to the plurality of first initial characteristic parameter sets according to the failed key equipment;
and determining a first characteristic parameter set corresponding to the current sampling moment according to the fuzzy weight matrix and the plurality of first initial characteristic parameter sets.
4. The method according to claim 3, wherein the determining the fuzzy weight matrix corresponding to the plurality of first initial feature parameter sets according to the failed key device comprises:
acquiring the corresponding relation between key equipment and the fuzzy weight matrix;
and determining fuzzy weight matrixes corresponding to the plurality of first initial characteristic parameter sets according to the corresponding relation.
5. The method according to claim 4, wherein before obtaining the correspondence between the key device and the fuzzy weight matrix, the method further comprises:
determining a fuzzy weight matrix corresponding to each key device;
and pre-establishing a corresponding relation between the key equipment and the fuzzy weight matrix.
6. The method according to claim 5, wherein the determining the fuzzy weight matrix corresponding to each of the key devices comprises:
determining a fault category corresponding to each first initial characteristic parameter set in a plurality of first initial characteristic parameter sets corresponding to each key device;
obtaining a first membership matrix of each fault category of each key device according to a first initial characteristic parameter set corresponding to each fault category of each key device;
normalizing the first membership degree corresponding to each characteristic parameter in the first membership degree matrix corresponding to all fault types of each key device to obtain the weight corresponding to each characteristic parameter of each fault type of each key device;
and combining the weights corresponding to the characteristic parameters of all fault types of each key device to obtain a fuzzy weight matrix corresponding to each key device.
7. The method of claim 6, wherein the obtaining a first membership matrix for each fault category of each key device according to the first initial characteristic parameter set corresponding to each fault category of each key device comprises:
determining the fault degree corresponding to each first initial characteristic parameter set in a plurality of first initial characteristic parameters corresponding to each fault category of each key device;
obtaining a second membership matrix of each fault degree of each fault category of each key device according to the first initial characteristic parameter set corresponding to each fault degree of each fault category of each key device;
averaging the second membership degrees corresponding to each characteristic parameter in the second membership degree matrix corresponding to each fault degree of each fault category of each key device to obtain a first membership degree corresponding to the characteristic parameter;
and combining the first membership corresponding to each characteristic parameter to obtain a first membership matrix corresponding to each fault type of each key device.
8. The method of claim 7, wherein obtaining a second membership matrix for each fault degree of each fault category of each key device according to the first initial feature parameter set corresponding to each fault degree of each fault category of each key device comprises:
determining a second membership degree corresponding to each characteristic parameter by using a normal function as a fuzzy operator;
and combining the second membership corresponding to each characteristic parameter to obtain the second membership matrix.
9. The method according to claim 8, wherein the determining the second degree of membership corresponding to each of the characteristic parameters by using the normal function as a fuzzy operator comprises:
counting each characteristic parameter in a plurality of characteristic parameter sets corresponding to different fault degrees of the same key equipment and the same fault type, and determining a normal function corresponding to each characteristic parameter;
and calculating a second membership degree corresponding to each characteristic parameter according to the value of each characteristic parameter.
10. A reliability prediction device for an overspeed protection system of a steam turbine, comprising:
a training module configured to train a recurrent neural network model for predicting reliability of a turbine overspeed protection system according to a plurality of first feature parameter sets of the system over a period of time;
an obtaining module configured to obtain a second initial feature parameter set of the system at a first time;
and the prediction module is configured to input the second initial characteristic parameter set to the trained recurrent neural network model to obtain the reliability of the system at a second moment.
CN201910955332.8A 2019-10-09 2019-10-09 Reliability prediction method and device for turbine overspeed protection system Active CN110717630B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910955332.8A CN110717630B (en) 2019-10-09 2019-10-09 Reliability prediction method and device for turbine overspeed protection system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910955332.8A CN110717630B (en) 2019-10-09 2019-10-09 Reliability prediction method and device for turbine overspeed protection system

Publications (2)

Publication Number Publication Date
CN110717630A true CN110717630A (en) 2020-01-21
CN110717630B CN110717630B (en) 2022-09-09

Family

ID=69212405

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910955332.8A Active CN110717630B (en) 2019-10-09 2019-10-09 Reliability prediction method and device for turbine overspeed protection system

Country Status (1)

Country Link
CN (1) CN110717630B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112948163A (en) * 2021-03-26 2021-06-11 中国航空无线电电子研究所 Method for evaluating influence of equipment on functional fault based on BP neural network

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070288409A1 (en) * 2005-05-31 2007-12-13 Honeywell International, Inc. Nonlinear neural network fault detection system and method
CN103033359A (en) * 2012-12-19 2013-04-10 西安交通大学 Wind turbine generator main gear fault diagnosis method of multi-feature multilevel comprehensive judgment
CN106125714A (en) * 2016-06-20 2016-11-16 南京工业大学 Failure Rate Forecasting Method in conjunction with BP neutral net Yu two parameters of Weibull
CN107992971A (en) * 2017-12-06 2018-05-04 中建材智慧工业科技有限公司 A kind of Forecasting Methodology and system in cement plant equipment and spare part service life
CN108090606A (en) * 2017-12-12 2018-05-29 上海应用技术大学 Equipment fault finds method and system
CN109492790A (en) * 2018-09-18 2019-03-19 北京光耀电力科技股份有限公司 Wind turbines health control method based on neural network and data mining
CN109858665A (en) * 2018-12-06 2019-06-07 国网河北省电力有限公司 Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070288409A1 (en) * 2005-05-31 2007-12-13 Honeywell International, Inc. Nonlinear neural network fault detection system and method
CN103033359A (en) * 2012-12-19 2013-04-10 西安交通大学 Wind turbine generator main gear fault diagnosis method of multi-feature multilevel comprehensive judgment
CN106125714A (en) * 2016-06-20 2016-11-16 南京工业大学 Failure Rate Forecasting Method in conjunction with BP neutral net Yu two parameters of Weibull
CN107992971A (en) * 2017-12-06 2018-05-04 中建材智慧工业科技有限公司 A kind of Forecasting Methodology and system in cement plant equipment and spare part service life
CN108090606A (en) * 2017-12-12 2018-05-29 上海应用技术大学 Equipment fault finds method and system
CN109492790A (en) * 2018-09-18 2019-03-19 北京光耀电力科技股份有限公司 Wind turbines health control method based on neural network and data mining
CN109858665A (en) * 2018-12-06 2019-06-07 国网河北省电力有限公司 Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
QIDAN ZHOU等: "Design of Fuzzy Neural Network Controller for Marine Steam Turbine Systerm", 《2008 FOURTH INTERNATIONAL CONFERENCE ON NATURAL COMPULATION》 *
于会荣: "基于模糊神经网络的矿井风机状态监测与故障诊断***研究", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅰ辑》 *
周磊等: "基于模糊SOM神经网络的汽轮机通流部分故障诊断研究", 《汽轮机技术》 *
梁为等: "《机械故障诊断》", 30 November 2015, 煤炭工业出版社 *
蒋超: "模糊神经网络在采煤机故障诊断中的应用", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅰ辑》 *
韩捷等: "《旋转机械故障机理及诊断技术》", 31 August 1997, 机械工业出版社 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112948163A (en) * 2021-03-26 2021-06-11 中国航空无线电电子研究所 Method for evaluating influence of equipment on functional fault based on BP neural network
CN112948163B (en) * 2021-03-26 2023-09-19 中国航空无线电电子研究所 Method for evaluating influence of equipment on functional failure based on BP neural network

Also Published As

Publication number Publication date
CN110717630B (en) 2022-09-09

Similar Documents

Publication Publication Date Title
Kong et al. Optimal sensor placement methodology of hydraulic control system for fault diagnosis
CN112749509B (en) Intelligent substation fault diagnosis method based on LSTM neural network
CN110311427A (en) The two stages N-K robust Fault of meter and probability of malfunction constrains Unit Combination method
CN113011248B (en) Nuclear power plant fault diagnosis method based on LSTM
CN113033078B (en) Construction method, system and early warning method of fault early warning model of relay protection equipment
CN104240781A (en) Signal distribution method and signal distribution system of digital instrument control system (DCS) of nuclear power plant
CN109740727A (en) A kind of hydraulic turbine shaft state monitoring method neural network based and system
CN110717630B (en) Reliability prediction method and device for turbine overspeed protection system
Ma et al. A measurement-simulation hybrid method for transient stability assessment and control based on the deviation energy
CN103344881A (en) Grid fault diagnosing method based on combined type cross entropy algorithm
CN114156901A (en) Method for detecting state abnormity of low-voltage distribution transformer reactive power compensation device
CN109932901A (en) A kind of two stages robust Unit Combination method counted and failure constrains
CN112489841A (en) Water level fault-tolerant control method for steam generator of nuclear power unit
Hagmar et al. On-line voltage instability prediction using an artificial neural network
Cilliers Benchmarking an expert fault detection and diagnostic system on the Three Mile Island accident event sequence
Seo et al. A study on modeling using big data and deep learning method for failure diagnosis of system
CN115693692A (en) Voltage qualification rate improving method based on power distribution network voltage data analysis
Batchu et al. Transient Stability Assessment and Enhancement in the Realm of Power System Dynamic Security Analysis: A Review and Scope of Methods
KR20010036423A (en) Method and apparatus for detecting fault using General Regression Neural Network
CN117273465B (en) Risk transfer method for cascade hydropower junction group
Bizjak et al. Vision 2020 dynamic security assessment in real time environment
Zhengganzhe et al. Risk forecast assessment of transmission line capacity increase under cyberattacks
Liu Knowledge-based systems in power systems: Applications and development methods
RU2297659C1 (en) Integrated system for automatic coordinated control of object
CN110601189B (en) Power grid safety evaluation method considering cascading trip

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant