CN117662491A - Main pump fault diagnosis method and system based on principal component and deep learning algorithm - Google Patents
Main pump fault diagnosis method and system based on principal component and deep learning algorithm Download PDFInfo
- Publication number
- CN117662491A CN117662491A CN202311834877.6A CN202311834877A CN117662491A CN 117662491 A CN117662491 A CN 117662491A CN 202311834877 A CN202311834877 A CN 202311834877A CN 117662491 A CN117662491 A CN 117662491A
- Authority
- CN
- China
- Prior art keywords
- main pump
- data
- real
- state
- operation data
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 91
- 238000003745 diagnosis Methods 0.000 title claims abstract description 64
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 36
- 238000013135 deep learning Methods 0.000 title claims abstract description 32
- 238000012544 monitoring process Methods 0.000 claims abstract description 95
- 238000004458 analytical method Methods 0.000 claims abstract description 29
- 230000002159 abnormal effect Effects 0.000 claims abstract description 26
- 230000008569 process Effects 0.000 claims abstract description 19
- 238000000513 principal component analysis Methods 0.000 claims description 20
- 238000013528 artificial neural network Methods 0.000 claims description 17
- 238000004590 computer program Methods 0.000 claims description 17
- 238000004364 calculation method Methods 0.000 claims description 12
- 125000004122 cyclic group Chemical group 0.000 claims description 12
- 238000004088 simulation Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 238000005311 autocorrelation function Methods 0.000 claims description 6
- 238000006073 displacement reaction Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 230000007246 mechanism Effects 0.000 claims description 6
- 238000001228 spectrum Methods 0.000 claims description 6
- 238000013112 stability test Methods 0.000 claims description 6
- 230000001186 cumulative effect Effects 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 230000005284 excitation Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000036961 partial effect Effects 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 7
- 238000012423 maintenance Methods 0.000 description 6
- 239000002826 coolant Substances 0.000 description 4
- 230000000306 recurrent effect Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000010354 integration Effects 0.000 description 3
- 230000002829 reductive effect Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000001373 regressive effect Effects 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000004992 fission Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000003449 preventive effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- 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
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Monitoring And Testing Of Nuclear Reactors (AREA)
Abstract
The invention discloses a main pump fault diagnosis method and a system based on a principal component and a deep learning algorithm, wherein the main pump fault diagnosis method comprises the following steps: acquiring real-time operation data of main pump equipment and storing the real-time operation data; according to real-time operation data of the main pump equipment, a main pump operation state circulation monitoring method based on main component analysis and multi-source data prediction is adopted to monitor the main pump state in the current moment and the subsequent preset time length of the main pump equipment, so as to obtain a monitoring state; the monitoring state comprises a normal state and an abnormal state; and according to the monitoring state and the real-time operation data of the main pump equipment, a time-frequency domain analysis method and an expert knowledge base are adopted, and the diagnosis and the discrimination of the typical fault mode are carried out by combining the process parameters acquired by the non-safety-level DCS system. The invention improves the reliability and accuracy of the on-line state monitoring and fault diagnosis work of the main pump, improves the intelligent level and reliability of the operation of the main pump equipment of the nuclear power plant, and reduces the economic loss of the power plant caused by the abnormal fault shutdown of the main pump equipment.
Description
Technical Field
The invention relates to the technical field of on-line state monitoring of reactor coolant pumps of a nuclear power plant, in particular to a main pump fault diagnosis method and system based on a main component and a deep learning algorithm.
Background
The reactor coolant pump (abbreviated as "main pump") is an important safety device for a nuclear power plant reactor and a primary loop system, and can ensure that heat generated by nuclear fission and decay in the reactor is continuously output to the secondary loop system, and is a part of the pressure boundary of the reactor coolant system, and the operation state of the reactor coolant pump is directly related to the performance and safety of the nuclear reactor system. However, as the service life of the equipment increases and the coupling between components affects, the main pump mechanical components face potential performance degradation and safety issues such as excessive vibration, bearing damage, mechanical wear, shaft seal leakage, and the like. According to the statistical analysis of the accidents of the nuclear power plant of the pressurized water reactor of the United states by Combustion Engineering company, the annual average shutdown maintenance time period caused by the failure of the main pump component is about 200.6 hours, which contributes to 1.67 percent of the unavailable time period of the nuclear power plant, and on average, 0.19 emergency shutdown is caused each year. Accordingly, it is desirable to take measures to understand and monitor and evaluate the health of the main pump equipment to improve the overall availability of nuclear power plant operation.
However, the operation and maintenance of the main pump equipment of the existing nuclear power plant mainly depend on a manual mode, the state monitoring of the key equipment of the nuclear power plant mainly depends on the threshold value alarm of the set on-line sensor, the effective analysis and integration of the multi-source monitoring sensing information are not realized, the discovery and early warning of early abnormal states cannot be carried out, the operation reliability of the key equipment such as the main pump is poor, the diagnosis after the faults are mainly judged by expert knowledge based on experience, the subjectivity is strong, the fault investigation efficiency is low, the real-time performance of the on-line state monitoring and fault diagnosis work of the main pump is poor, the reliability is low and the accuracy is low.
In view of this, the present application is specifically proposed.
Disclosure of Invention
The technical problem to be solved by the invention is that the existing main pump state monitoring method mainly depends on a manual mode, depends on the threshold value alarm of the set on-line sensor, does not realize the effective analysis and integration of multisource monitoring sensing information, cannot discover and early warn the early abnormal state, has poor operation reliability of key equipment such as a main pump and the like, and has strong subjectivity and low fault checking efficiency due to the fact that diagnosis after faults occur mainly depends on expert knowledge judgment based on experience, and has poor real-time performance, low reliability and low accuracy of the on-line state monitoring and fault diagnosis work of the main pump.
The invention aims to provide a main pump fault diagnosis method and a main pump fault diagnosis system based on a principal component and a deep learning algorithm, which can further improve the reliability and the accuracy of the on-line state monitoring and fault diagnosis work of a main pump, improve the intelligent level and the reliability of the operation of main pump equipment of a nuclear power plant, and reduce the economic loss of the power plant caused by the abnormal fault shutdown of the main pump equipment.
The invention is realized by the following technical scheme:
in a first aspect, the present invention provides a main pump failure diagnosis method based on a principal component and a deep learning algorithm, the method comprising:
acquiring real-time operation data of main pump equipment, and storing the real-time operation data of the main pump equipment;
according to real-time operation data of the main pump equipment, a main pump operation state circulation monitoring method based on main component analysis and multi-source data prediction is adopted to monitor the main pump state in the current moment and the subsequent preset time length of the main pump equipment, so as to obtain a monitoring state; the monitoring state comprises a normal state and an abnormal state;
and according to the monitoring state and the real-time operation data of the main pump equipment, a time-frequency domain analysis method and an expert knowledge base are adopted, and the diagnosis and the discrimination of the typical fault mode are carried out by combining the process parameters acquired by the non-safety-level DCS system.
Further, the real-time operation data of the main pump device refers to the real-time operation data of the sensor collected by the data interface of the main pump device itself;
the real-time operation data of the main pump equipment comprises first data acquired by a special high-frequency acquisition device and second data acquired by a non-safety-level DCS system; the first data includes raw digital signals of vibration and shaft displacement of the main pump device, and the second data includes signals of temperature, flow, rotation speed, current, pressure and the like of the main pump device.
Further, the main pump running state circulation monitoring method based on principal component analysis and multi-source data prediction comprises a main pump on-line state monitoring method based on principal component analysis, and the steps are as follows:
a1, forming an n multiplied by m data matrix by acquiring historical operation data under normal operation conditions of main pump equipment, wherein n is the number of samples, and m is the number of measuring points; calculating a mean value and a variance of the historical operation data, and performing standardization processing to form a standardized sample;
a2, obtaining a covariance matrix of a standardized sample through calculation, carrying out feature vector decomposition to obtain a feature value and a feature vector, and obtaining the number t of principal elements by adopting a cumulative variance percentage method; calculating projection of the feature vector on the feature space to finish T 2 And SPE two statistics index quantity control limit calculation;
step A3, an offline principal component model is established, and a process monitoring step is carried out by using actual equipment operation data: the real-time operation data vector at the ith moment is standardized to obtain a vector x i And vector x i Substituting the principal component model to obtain an estimated vector
Step A4, calculating a vector x i Statistics T of 2 And SPE, obtaining real-time statistics; and real-time statistics and offline mastersComparing the control limits determined by the meta model, and judging the monitoring state of the main pump equipment; and if the real-time statistic exceeds the control limit, indicating that the monitored main pump equipment is abnormal, otherwise, indicating that the monitored main pump equipment is normally operated.
Further, the main pump running state circulation monitoring method based on principal component analysis and multi-source data prediction also comprises multi-source parameter prediction based on autoregressive moving average, and the steps are as follows:
step B1, preprocessing historical operation data to obtain preprocessed historical operation data;
step B2, carrying out stability test on the preprocessed historical operation data, and calculating an autocorrelation function and a partial autocorrelation function of the preprocessed historical operation data;
step B3, combining stability test selection, and determining relevant parameters based on an autoregressive moving average model;
and step B4, inputting the real-time operation data into a trained autoregressive moving average model to predict, so as to achieve a real-time data prediction result.
Further, the main pump running state circulation monitoring method based on principal component analysis and multi-source data prediction also comprises multi-source parameter prediction based on a circulation neural network GRU, and the steps are as follows:
step C1, dividing a data set formed by historical operation data into a training set and a testing set, and normalizing and preprocessing the data set by using a MinMaxScaler estimator to obtain processed training data and testing data;
step C2, inputting the processed training data into a cyclic neural network for training, and storing the existing weight and bias in the cyclic neural network after the loss function reaches a preset standard;
step C3, inputting the processed test data into a cyclic neural network for feedforward calculation to obtain a predicted output; and performing inverse normalization on the prediction output by using a MinMaxScaler estimator to obtain a prediction result.
Further, according to the monitoring state and the real-time operation data of the main pump equipment, a time-frequency domain analysis method and an expert knowledge base are adopted, and the diagnosis and the discrimination of the typical fault mode are carried out by combining the process parameters acquired by the non-safety-level DCS system, including:
step D1, according to the monitoring state, constructing simulation dynamics models under different fault modes by analyzing fault mechanisms of unbalanced faults, rub-impact faults and base loosening faults of a rotor bearing system;
step D2, calculating excitation response under typical faults such as vibration and the like by solving simulation dynamics models under different fault modes, so as to obtain a simulation result;
step D3, analyzing time domain waveforms, axis tracks and vibration characteristic values through simulation calculation results of vibration and the like, and establishing a fault expert knowledge base under a typical fault mode of main pump equipment;
step D4, obtaining a time domain waveform, an axis track and a spectrogram of the real-time operation signal by carrying out data preprocessing on the real-time operation signal, and analyzing the characteristic frequency of the time domain waveform, the axis track and the spectrogram to obtain a related frequency spectrum of the real-time operation signal; the relevant frequency spectrum of the real-time operation signal and the data characteristics of the second data acquired by the corresponding non-safety level DCS system at the same time are compared and analyzed with a fault expert knowledge base, so that diagnosis and discrimination of a typical fault mode of the main pump are realized;
the process parameters collected by the non-safety-level DCS system comprise high-pressure leakage flow, shaft seal outlet pressure, bearing temperature, motor current, rotating speed and the like.
In a second aspect, the present invention further provides a main pump fault diagnosis system based on a principal component and a deep learning algorithm, the system using the above-described main pump fault diagnosis method based on the principal component and the deep learning algorithm; the system comprises:
the device comprises an acquisition unit, a control unit and a control unit, wherein the acquisition unit is used for acquiring real-time operation data of main pump equipment and storing the real-time operation data of the main pump equipment;
the state monitoring unit is used for monitoring the state of the main pump in the current moment and the subsequent preset time length of the main pump equipment by adopting a main pump operation state circulation monitoring method based on main component analysis and multi-source data prediction according to the real-time operation data of the main pump equipment to obtain a monitored state; the monitoring state comprises a normal state and an abnormal state;
and the fault diagnosis unit is used for diagnosing and judging a typical fault mode by adopting a time-frequency domain analysis method and an expert knowledge base according to the monitoring state and real-time operation data of the main pump equipment and combining process parameters acquired by the non-safety-level DCS system.
Further, the real-time operation data of the main pump device refers to the real-time operation data of the sensor collected by the data interface of the main pump device itself;
the real-time operation data of the main pump equipment comprises first data acquired by a special high-frequency acquisition device and second data acquired by a non-safety-level DCS system; the first data includes raw digital signals of vibration and shaft displacement of the main pump device, and the second data includes signals of temperature, flow, rotation speed, current, pressure and the like of the main pump device.
In a third aspect, the present invention further provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor implementing the above-described main pump fault diagnosis method based on a principal component and a deep learning algorithm when executing the computer program.
In a fourth aspect, the present invention further provides a computer readable storage medium storing a computer program, which when executed by a processor, implements the above-described main pump fault diagnosis method based on a principal component and a deep learning algorithm.
Compared with the prior art, the invention has the following advantages and beneficial effects:
the invention provides a main component and deep learning algorithm-based main pump fault diagnosis method and system, and provides an integrated data-driven algorithm framework to realize on-line state monitoring of main pump equipment.
(1) Aiming at the problems of multi-source sensing monitoring and monitoring data time sequence development of main pump equipment, the invention provides a main pump running state circulation monitoring method based on principal component analysis and multi-source data prediction, which can realize main pump state monitoring at the current moment and in the subsequent specific time length of main pump equipment.
(2) By combining expert knowledge of a typical failure mechanism of a main pump with a time-frequency domain analysis method, the distinction and the discrimination of the typical failure mode after the main pump is abnormal in operation can be realized, and more accurate technical support is provided for subsequent operation and maintenance decisions of equipment.
(3) The integrated framework model based on the data driving method meets the monitoring and diagnosis requirements of the main pump equipment at different moments more accurately, and compared with a single monitoring and diagnosis model, the early warning of the abnormal state of the main pump can be realized earlier. The method can further improve the reliability and accuracy of the on-line state monitoring and fault diagnosis work of the main pump, and has important economic significance for improving the operation reliability of the main pump equipment of the nuclear power plant and reducing the economic loss of the power plant caused by the abnormal fault shutdown of the main pump equipment.
Drawings
The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and are incorporated in and constitute a part of this application, illustrate embodiments of the invention. In the drawings:
FIG. 1 is a flow chart of a main pump fault diagnosis method based on a principal component and a deep learning algorithm;
FIG. 2 is a detailed flow chart of a main pump fault diagnosis method based on a principal component and a deep learning algorithm;
FIG. 3 is a flow chart of a main pump on-line state monitoring method based on principal component analysis according to the present invention;
FIG. 4 is a flow chart of the multi-source parameter prediction based on autoregressive moving average (ARIMA) of the present invention;
FIG. 5 is a flow chart of the multi-source parameter prediction based on the recurrent neural network GRU of the invention;
FIG. 6 is a flow chart of fault diagnosis based on time-frequency domain analysis and expert knowledge base according to the present invention;
FIG. 7 is a block diagram of a main pump fault diagnosis system based on principal components and a deep learning algorithm.
Detailed Description
For the purpose of making apparent the objects, technical solutions and advantages of the present invention, the present invention will be further described in detail with reference to the following examples and the accompanying drawings, wherein the exemplary embodiments of the present invention and the descriptions thereof are for illustrating the present invention only and are not to be construed as limiting the present invention.
The existing main pump state monitoring method mainly depends on a manual mode, depends on the set threshold value of an online sensor for alarming, does not realize effective analysis and integration of multisource monitoring sensing information, cannot find and early warn early abnormal states, is poor in operation reliability of key equipment such as a main pump, and is also mainly judged by expert knowledge based on experience after fault, and is high in subjectivity, low in fault detection efficiency, poor in real-time performance, low in reliability and low in accuracy of main pump online state monitoring and fault diagnosis work.
The invention designs a main component and deep learning algorithm-based main pump fault diagnosis method and system, namely an integrated data-driven algorithm framework is provided to realize on-line state monitoring of main pump equipment, so that the reliability and the accuracy of on-line state monitoring and fault diagnosis work of the main pump can be further improved, the intelligent level and the reliability of operation of main pump equipment of a nuclear power plant are improved, the possibility of economic loss of the power plant caused by abnormal fault shutdown of key equipment such as the main pump is reduced, and a follow-up 'unattended' digital nuclear power plant and intelligent nuclear power plant implementation foundation is laid.
Specifically, the invention designs an integrated data-driven-based flow framework for monitoring the on-line state of the main pump equipment by taking the main pump equipment of the nuclear power plant as a research object. The method framework integrates a dimension reduction method such as principal component analysis (Principal Component Analysis, PCA), a data prediction method such as autoregressive moving average (Auto Regressive Integrated Moving Average, ARIMA) and cyclic neural network (Gated Recurrent Unit, GRU), and an expert knowledge and time-frequency domain analysis method. The effective analysis of the multi-source monitoring data of the main pump can be completed through the dimension reduction methods such as PCA and the like, and the real-time monitoring of the operation state of the main pump is realized. Meanwhile, based on prediction models such as ARIMA and the like, parameter constraint relation of process signals is extracted, abnormal operation behaviors are captured timely, false alarm rate is reduced, and a foundation is laid for subsequent main pump fault diagnosis based on expert knowledge and a time-frequency domain analysis method. It is worth emphasizing that the research work in the invention has both basic scientific research value and practical application prospect. From the scientific significance level, the invention relates to the field of interdiscipling of expert knowledge and artificial intelligence technology of typical failure mechanisms of a main pump, and is based on researching abnormal monitoring and failure mode distinction of the typical failure modes of the main pump. From the technical application point of view, the method is based on the following characteristics and advantages:
(1) Aiming at the problems of multi-source sensing monitoring and monitoring data time sequence development of main pump equipment, the invention provides a main pump running state circulation monitoring method based on principal component analysis and multi-source data prediction, which can realize main pump state monitoring at the current moment and in the subsequent specific time length of main pump equipment.
(2) By combining expert knowledge of a typical failure mechanism of a main pump with a time-frequency domain analysis method, the distinction and the discrimination of the typical failure mode after the main pump is abnormal in operation can be realized, and more accurate technical support is provided for subsequent operation and maintenance decisions of equipment.
(3) The integrated framework model based on the data driving method meets the monitoring and diagnosis requirements of the main pump equipment at different moments more accurately, and compared with a single monitoring and diagnosis model, the early warning of the abnormal state of the main pump can be realized earlier. The method can further improve the reliability and accuracy of the on-line state monitoring and fault diagnosis work of the main pump, and has important economic significance for improving the operation reliability of the main pump equipment of the nuclear power plant and reducing the economic loss of the power plant caused by the abnormal fault shutdown of the main pump equipment.
Example 1
As shown in fig. 1, the main pump fault diagnosis method based on the principal component and the deep learning algorithm of the present invention comprises:
step 1, acquiring real-time operation data of main pump equipment, and storing the real-time operation data of the main pump equipment;
in this embodiment, the real-time operation data of the main pump device refers to the real-time operation data of the sensor collected by using the data interface of the main pump device itself;
the real-time operation data of the main pump equipment comprises first data acquired by a special high-frequency acquisition device and second data acquired by a non-safety-level DCS system; the first data includes raw digital signals of vibration and shaft displacement of the main pump device, and the second data includes signals of temperature, flow, rotation speed, current, pressure and the like of the main pump device.
In this embodiment, the obtained real-time operation data of the main pump device is saved in the database by the data storage software of the system for other modules to call.
Step 2, monitoring the state of the main pump in the current moment and the subsequent preset time length of the main pump equipment by adopting a main pump operation state circulation monitoring method based on main component analysis and multi-source data prediction according to real-time operation data of the main pump equipment to obtain a monitoring state; the monitoring state comprises a normal state and an abnormal state;
the main pump running state circulation monitoring method based on principal component analysis and multi-source data prediction in the step 2 comprises online state monitoring and multi-source data prediction;
(1) Main pump on-line state monitoring method based on principal component analysis PCA: and the abnormal detection and fault alarm of the equipment are realized by utilizing the real-time operation data of the main pump equipment and combining PCA, filtering, feature extraction and the like.
(2) Multi-source parameter prediction: trend prediction within normal thresholds of monitoring parameters of equipment is achieved by adopting autoregressive moving averages (Auto Regressive Integrated Moving Average, ARIMA) and artificial neural networks, and an important engineering reference basis is provided for the transition of the equipment from planned maintenance to preventive maintenance.
(1) A main pump on-line state monitoring method based on principal component analysis. As shown in fig. 3, fig. 3 is a flow chart of a main pump on-line state monitoring method based on principal component analysis; the main pump on-line state monitoring method based on the principal component analysis comprises the following steps:
a1, forming an n multiplied by m data matrix by acquiring historical operation data under normal operation conditions of main pump equipment, wherein n is the number of samples, and m is the number of measuring points; because the process parameters measured by the sensors used for measuring points comprise temperature, pressure, flow and the like, the acquired historical operation data set is multi-source heterogeneous data, and in order to avoid adverse effects caused by different data magnitude and dimension, the average value and variance of the historical operation data are calculated, and standardized processing is carried out to form a standardized sample;
a2, obtaining a covariance matrix of a standardized sample through calculation, carrying out feature vector decomposition to obtain a feature value and a feature vector, and obtaining the number t of principal elements by adopting a cumulative variance percentage method; calculating projection of the feature vector on the feature space to finish T 2 And SPE two statistics index quantity control limit calculation;
step A3, an offline principal component model is established, and a process monitoring step is carried out by using actual equipment operation data: the real-time operation data vector at the ith moment is standardized to obtain a vector x i And vector x i Substituting the principal component model to obtain an estimated vector
Step A4, calculating a vector x i Statistics T of 2 And SPE, obtaining real-time statistics; comparing the real-time statistic with a control limit determined by an offline principal component model, and judging the monitoring state of main pump equipment; and if the real-time statistic exceeds the control limit, indicating that the monitored main pump equipment is abnormal, otherwise, indicating that the monitored main pump equipment is normally operated.
(2) Multi-source parameter prediction
(2-1) multi-source parameter prediction based on autoregressive moving average. As shown in fig. 4, fig. 4 is a flowchart of multi-source parameter prediction based on autoregressive moving average (ARIMA), and the steps are as follows:
step B1, preprocessing historical operation data to obtain preprocessed historical operation data;
step B2, carrying out stability test on the preprocessed historical operation data, and calculating an autocorrelation function and a partial autocorrelation function of the preprocessed historical operation data;
step B3, combining with stability test selection, determining (p, i, q) related parameters in an autoregressive moving average model;
and step B4, inputting the real-time operation data into a trained autoregressive moving average model to predict, so as to achieve a real-time data prediction result.
(2-2) Multi-source parameter prediction based on recurrent neural network GRU As shown in FIG. 5, FIG. 5 is a flow chart of Multi-source parameter prediction based on recurrent neural network GRU, comprising the steps of:
step C1, dividing a data set formed by historical operation data into a training set and a testing set, and normalizing and preprocessing the data set by using a MinMaxScaler estimator to obtain processed training data and testing data;
step C2, inputting the processed training data into a cyclic neural network for training, and storing the existing weight and bias in the cyclic neural network after the loss function reaches a preset standard;
step C3, inputting the processed test data into a cyclic neural network for feedforward calculation to obtain a predicted output; and performing inverse normalization on the prediction output by using a MinMaxScaler estimator to obtain a prediction result.
And 3, according to the monitoring state and the real-time operation data of the main pump equipment, adopting a time-frequency domain analysis method and an expert knowledge base, and carrying out diagnosis and discrimination of a typical fault mode by combining process parameters acquired by the non-safety-level DCS system.
As shown in fig. 6, fig. 6 is a fault diagnosis flowchart based on a time-frequency domain analysis method and an expert knowledge base, and step 3 specifically includes:
step D1, according to the monitoring state, constructing simulation dynamics models under different fault modes by analyzing fault mechanisms of unbalanced faults, rub-impact faults and base loosening faults of a rotor bearing system;
step D2, calculating excitation response under typical faults such as vibration and the like by solving simulation dynamics models under different fault modes, so as to obtain a simulation result;
step D3, analyzing time domain waveforms, axis tracks and vibration characteristic values through simulation calculation results of vibration and the like, and establishing a fault expert knowledge base under a typical fault mode of main pump equipment;
step D4, obtaining a time domain waveform, an axis track and a spectrogram of the real-time operation signal by carrying out data preprocessing on the real-time operation signal, and analyzing the characteristic frequency of the time domain waveform, the axis track and the spectrogram to obtain a related frequency spectrum of the real-time operation signal; the relevant frequency spectrum of the real-time operation signal and the data characteristics of the second data acquired by the corresponding non-safety level DCS system at the same time are compared and analyzed with a fault expert knowledge base, so that diagnosis and discrimination of a typical fault mode of the main pump are realized;
the process parameters collected by the non-safety-level DCS system comprise high-pressure leakage flow, shaft seal outlet pressure, bearing temperature, motor current, rotating speed and the like.
In the technical scheme, in the step 3, the type, the position and the cause of the equipment fault are judged by utilizing expert knowledge of the main pump equipment and combining algorithms such as a time-frequency domain analysis method, an axis track model and the like to identify a typical fault mode of the main pump equipment.
The method adopts an integrated framework to carry out on-line monitoring of the state of the main pump, and comprises a main pump on-line state monitoring method based on principal component analysis PCA, multi-source parameter prediction based on autoregressive moving average, multi-source parameter prediction based on a cyclic neural network GRU, a time-frequency domain analysis based method and the like. The effective monitoring of vibration signals is realized through a principal component analysis PCA technology, trend prediction is carried out on various signals of the main pump within a normal threshold based on an ARIMA model and a GRU algorithm, and the false alarm rate is reduced. And finally, diagnosing and analyzing the typical failure mode of the main pump based on the equipment expert knowledge system and the time-frequency domain analysis technology.
The method can improve the accuracy of the fault diagnosis result and the trend prediction result, further improve the operation reliability of the main pump equipment of the nuclear power plant, reduce the economic loss of the power plant caused by the abnormal fault shutdown of the main pump equipment, improve the economy of the power plant, and have important economic significance and wide application prospect.
Example 2
As shown in fig. 7, the present embodiment is different from embodiment 1 in that the present embodiment further provides a main pump failure diagnosis system based on a principal component and a deep learning algorithm, which uses the main pump failure diagnosis method of embodiment 1 based on the principal component and the deep learning algorithm; the system comprises:
the device comprises an acquisition unit, a control unit and a control unit, wherein the acquisition unit is used for acquiring real-time operation data of main pump equipment and storing the real-time operation data of the main pump equipment;
the state monitoring unit is used for monitoring the state of the main pump in the current moment and the subsequent preset time length of the main pump equipment by adopting a main pump operation state circulation monitoring method based on main component analysis and multi-source data prediction according to the real-time operation data of the main pump equipment to obtain a monitored state; the monitoring state comprises a normal state and an abnormal state;
and the fault diagnosis unit is used for diagnosing and judging a typical fault mode by adopting a time-frequency domain analysis method and an expert knowledge base according to the monitoring state and real-time operation data of the main pump equipment and combining process parameters acquired by the non-safety-level DCS system.
As a further implementation, the real-time operation data of the main pump device refers to the real-time operation data of the collecting sensor by using the data interface of the main pump device itself;
the real-time operation data of the main pump equipment comprises first data acquired by a special high-frequency acquisition device and second data acquired by a non-safety-level DCS system; the first data includes raw digital signals of vibration and shaft displacement of the main pump device, and the second data includes signals of temperature, flow, rotation speed, current, pressure and the like of the main pump device.
The execution process of each unit is performed according to the main pump fault diagnosis method based on the principal component and the deep learning algorithm in embodiment 1, and the detailed description is omitted in this embodiment.
Meanwhile, the invention also provides a computer device, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to realize the main pump fault diagnosis method based on the principal component and the deep learning algorithm in the embodiment 1.
Meanwhile, the present invention also provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the main pump failure diagnosis method based on the principal component and the deep learning algorithm of embodiment 1.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the scope of the invention, but to limit the invention to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (10)
1. The main pump fault diagnosis method based on the principal component and the deep learning algorithm is characterized by comprising the following steps:
acquiring real-time operation data of main pump equipment, and storing the real-time operation data of the main pump equipment;
according to the real-time operation data of the main pump equipment, a main pump operation state circulation monitoring method based on main component analysis and multi-source data prediction is adopted to monitor the main pump state in the current moment and the subsequent preset time length of the main pump equipment, so as to obtain a monitoring state; the monitoring state comprises a normal state and an abnormal state;
and according to the monitoring state and the real-time operation data of the main pump equipment, a time-frequency domain analysis method and an expert knowledge base are adopted, and the diagnosis and the discrimination of the typical fault mode are carried out by combining the process parameters acquired by the non-safety-level DCS system.
2. The main pump failure diagnosis method based on the principal component and the deep learning algorithm according to claim 1, wherein the real-time operation data of the main pump device refers to the real-time operation data of the acquisition sensor using the data interface of the main pump device itself;
the real-time operation data of the main pump equipment comprises first data acquired by a special high-frequency acquisition device and second data acquired by a non-safety-level DCS system; the first data includes raw digital signals of vibration and shaft displacement of the main pump device, and the second data includes temperature, flow, rotational speed, current and pressure signals of the main pump device.
3. The main pump fault diagnosis method based on principal component and deep learning algorithm according to claim 1, wherein the main pump operation state cycle monitoring method based on principal component analysis and multi-source data prediction comprises a main pump on-line state monitoring method based on principal component analysis, comprising the steps of:
a1, forming an n multiplied by m data matrix by acquiring historical operation data under normal operation conditions of main pump equipment, wherein n is the number of samples, and m is the number of measuring points; calculating a mean value and a variance of the historical operation data, and performing standardization processing to form a standardized sample;
a2, obtaining a covariance matrix of the standardized sample through calculation, carrying out feature vector decomposition to obtain a feature value and a feature vector, and obtaining the number t of principal elements by adopting a cumulative variance percentage method; calculating the projection of the feature vector on the feature space to finish T 2 And SPE two statistics index quantity control limit calculation;
step A3, an offline principal component model is established, and a process monitoring step is carried out by using actual equipment operation data: the real-time operation data vector at the ith moment is standardized to obtain a vector x i And the vector x i Substituting the principal component model to obtain an estimated vector
Step A4, calculating the vector x i Statistics T of 2 And SPE, obtaining real-time statistics; comparing the real-time statistic with a control limit determined by an offline principal component model, and judging the monitoring state of main pump equipment; if the fact is trueAnd if the time statistics exceeds the control limit, the monitored main pump equipment is abnormal, and if the time statistics exceeds the control limit, the monitored main pump equipment is normal.
4. The main pump fault diagnosis method based on principal component and deep learning algorithm according to claim 3, wherein the main pump operation state cycle monitoring method based on principal component analysis and multi-source data prediction further comprises multi-source parameter prediction based on auto-regressive moving average, comprising the steps of:
step B1, preprocessing historical operation data to obtain preprocessed historical operation data;
step B2, carrying out stability test on the preprocessed historical operation data, and calculating an autocorrelation function and a partial autocorrelation function of the preprocessed historical operation data;
step B3, combining stability test selection, and determining relevant parameters based on an autoregressive moving average model;
and step B4, inputting the real-time operation data into a trained autoregressive moving average model to predict, so as to achieve a real-time data prediction result.
5. The main pump fault diagnosis method based on principal component and deep learning algorithm according to claim 3, wherein the main pump operation state cycle monitoring method based on principal component analysis and multi-source data prediction further comprises multi-source parameter prediction based on a cyclic neural network, comprising the steps of:
step C1, dividing a data set formed by historical operation data into a training set and a testing set, normalizing and preprocessing the data set to obtain processed training data and testing data;
step C2, inputting the processed training data into a cyclic neural network for training, and storing the existing weight and bias in the cyclic neural network after the loss function reaches a preset standard;
step C3, inputting the processed test data into a cyclic neural network for feedforward calculation to obtain a predicted output; and performing inverse normalization on the prediction output to obtain a prediction result.
6. The main pump fault diagnosis method based on the principal component and the deep learning algorithm according to claim 1, wherein the diagnosis and discrimination of the typical fault mode are performed by using a time-frequency domain analysis method and an expert knowledge base according to the monitoring state and real-time operation data of the main pump equipment, in combination with process parameters acquired by a non-safety level DCS system, comprising:
step D1, according to the monitoring state, constructing simulation dynamics models under different fault modes by analyzing fault mechanisms of unbalanced faults, rub-impact faults and base loosening faults of a rotor bearing system;
step D2, calculating excitation response under typical faults by solving simulation dynamics models under different fault modes to obtain simulation results;
step D3, a fault expert knowledge base under a typical fault mode of the main pump equipment is established by analyzing the time domain waveform, the axis locus and the vibration characteristic value of the simulation calculation result;
step D4, obtaining a time domain waveform, an axis track and a spectrogram of the real-time operation signal by carrying out data preprocessing on the real-time operation signal, and analyzing the characteristic frequency of the time domain waveform, the axis track and the spectrogram to obtain a related frequency spectrum of the real-time operation signal; comparing and analyzing the related frequency spectrum of the real-time operation signal and the data characteristics of the second data acquired by the corresponding non-safety level DCS system at the same time with the fault expert knowledge base to realize diagnosis and judgment of a typical fault mode of the main pump;
the process parameters collected by the non-safety-level DCS system comprise high-pressure leakage flow, shaft seal outlet pressure, bearing temperature, motor current and rotating speed.
7. A main pump failure diagnosis system based on a principal component and a deep learning algorithm, characterized in that the system uses the main pump failure diagnosis method based on a principal component and a deep learning algorithm as set forth in any one of claims 1 to 6; the system comprises:
the device comprises an acquisition unit, a control unit and a control unit, wherein the acquisition unit is used for acquiring real-time operation data of main pump equipment and storing the real-time operation data of the main pump equipment;
the state monitoring unit is used for monitoring the state of the main pump in the current moment and the subsequent preset time length of the main pump equipment by adopting a main pump operation state circulation monitoring method based on main component analysis and multi-source data prediction according to the real-time operation data of the main pump equipment to obtain a monitored state; the monitoring state comprises a normal state and an abnormal state;
and the fault diagnosis unit is used for diagnosing and judging a typical fault mode by adopting a time-frequency domain analysis method and an expert knowledge base according to the monitoring state and the real-time operation data of the main pump equipment and combining process parameters acquired by the non-safety-level DCS system.
8. The main pump failure diagnosis system based on the principal component and the deep learning algorithm according to claim 7, wherein the real-time operation data of the main pump apparatus refers to real-time operation data of a collection sensor using a data interface of the main pump apparatus itself;
the real-time operation data of the main pump equipment comprises first data acquired by a special high-frequency acquisition device and second data acquired by a non-safety-level DCS system; the first data includes raw digital signals of vibration and shaft displacement of the main pump device, and the second data includes temperature, flow, rotational speed, current and pressure signals of the main pump device.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the main pump failure diagnosis method based on the principal component and the deep learning algorithm as claimed in any one of claims 1 to 6 when executing the computer program.
10. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the main pump failure diagnosis method based on the principal component and the deep learning algorithm as claimed in any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311834877.6A CN117662491A (en) | 2023-12-28 | 2023-12-28 | Main pump fault diagnosis method and system based on principal component and deep learning algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311834877.6A CN117662491A (en) | 2023-12-28 | 2023-12-28 | Main pump fault diagnosis method and system based on principal component and deep learning algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117662491A true CN117662491A (en) | 2024-03-08 |
Family
ID=90071469
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311834877.6A Pending CN117662491A (en) | 2023-12-28 | 2023-12-28 | Main pump fault diagnosis method and system based on principal component and deep learning algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117662491A (en) |
-
2023
- 2023-12-28 CN CN202311834877.6A patent/CN117662491A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109524139B (en) | Real-time equipment performance monitoring method based on equipment working condition change | |
| CN110414155B (en) | Fan component temperature abnormity detection and alarm method with single measuring point | |
| CN112784373B (en) | Fault early warning method for wind turbine generator gearbox | |
| US6587737B2 (en) | Method for the monitoring of a plant | |
| CN110701137B (en) | Intelligent online detection and diagnosis device and method for hydraulic system of heading machine | |
| CN108803569A (en) | Station boiler diagnostic expert system and its method for diagnosing faults | |
| CN116320832B (en) | Monitoring equipment fault monitoring method and device | |
| CN115559890A (en) | Water pump unit operation fault prediction and adjustment method and system | |
| CN110469496B (en) | Intelligent early warning method and system for water pump | |
| CN111098463A (en) | Injection molding machine fault diagnosis system and diagnosis method | |
| CN113705924A (en) | Intelligent diagnosis method and system for thermal control equipment | |
| CN114841580A (en) | Generator fault detection method based on hybrid attention mechanism | |
| CN112926656A (en) | Method, system and equipment for predicting state of circulating water pump of nuclear power plant | |
| CN112124229A (en) | Remote real-time fault diagnosis method for emergency pump truck | |
| CN111486043A (en) | Lower rack fault diagnosis method based on hydro-turbo generator set runout data | |
| CN117786461A (en) | Water pump fault diagnosis method, control device and storage medium thereof | |
| CN114738132A (en) | Supercharger speed sensor fault diagnosis method and system based on real ship data | |
| CN112016193B (en) | Online prediction method and system for lubrication failure of shield tunneling machine system | |
| CN117350377A (en) | Knowledge graph driving-based equipment fault diagnosis method and device | |
| CN112308391A (en) | Real-time monitoring and anomaly detection method for equipment state based on neural network | |
| CN113313365A (en) | Degradation early warning method and device for primary air fan | |
| US11339763B2 (en) | Method for windmill farm monitoring | |
| CN117662491A (en) | Main pump fault diagnosis method and system based on principal component and deep learning algorithm | |
| Sun et al. | Anomaly detection of marine diesel engine valve system based on Mahalanobis distance | |
| CN117034157B (en) | Hydropower equipment fault identification method and system combining multimodal operation data |
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 |