CN114201825A - Method and system for evaluating equipment performance degradation state based on combination characteristics - Google Patents

Method and system for evaluating equipment performance degradation state based on combination characteristics Download PDF

Info

Publication number
CN114201825A
CN114201825A CN202111316146.3A CN202111316146A CN114201825A CN 114201825 A CN114201825 A CN 114201825A CN 202111316146 A CN202111316146 A CN 202111316146A CN 114201825 A CN114201825 A CN 114201825A
Authority
CN
China
Prior art keywords
data
evaluation
wavelet
entropy
equipment
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
Application number
CN202111316146.3A
Other languages
Chinese (zh)
Inventor
张燧
胡雪琛
曾谁飞
王青天
李小翔
刘跃
陈朝晖
张开开
冯帆
陈沐新
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.)
Huaneng Clean Energy Research Institute
Clean Energy Branch of Huaneng Zhejiang Energy Development Co Ltd
Original Assignee
Huaneng Clean Energy Research Institute
Clean Energy Branch of Huaneng Zhejiang Energy Development 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 Huaneng Clean Energy Research Institute, Clean Energy Branch of Huaneng Zhejiang Energy Development Co Ltd filed Critical Huaneng Clean Energy Research Institute
Priority to CN202111316146.3A priority Critical patent/CN114201825A/en
Publication of CN114201825A publication Critical patent/CN114201825A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/17—Mechanical parametric or variational design
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/04—Ageing analysis or optimisation against ageing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching

Abstract

The application provides an evaluation method and system of equipment performance degradation state based on combined characteristics, wherein the method comprises the following steps: acquiring evaluation data of at least one device in the new energy system; decomposing the evaluation data through Wavelet Packet Decomposition (WPD), and reconstructing corresponding wavelet data; calculating the entropy of each wavelet data according to the energy of each wavelet data, decomposing and converting the entropy of each wavelet data through EMD (empirical mode decomposition), and extracting a combined feature vector corresponding to the evaluation data; inputting the combined feature vector to a trained self-organizing map (SOM) network, and calculating a health evaluation index; and constructing an RUL model of the residual service life, and predicting the residual service time of the equipment according to the health evaluation curve through the RUL model. The method can effectively identify the current degradation stage of the equipment, accurately reflect the current health state of the equipment, facilitate early warning, reduce maintenance cost and improve the safety of the equipment.

Description

Method and system for evaluating equipment performance degradation state based on combination characteristics
Technical Field
The present application relates to the field of device management technologies, and in particular, to a method and a system for evaluating a device performance degradation state based on a combination feature.
Background
With the development of new energy technology, the popularity of new energy systems is gradually increasing. The new energy system is formed by connecting various devices and a pipe network, has the input and conversion of various energy sources, and can supply various energy sources to different users. Various types of equipment can be included in the new energy system, and for example, photovoltaic equipment, ground source heat pumps, wind energy equipment, energy storage equipment and the like can be included. Since the health of a large number of devices in the integrated energy system may be damaged due to the influence of long-term operation, environmental changes, frequent start and stop, and the like, and even when the regular maintenance time is not reached, the devices may be out of order, which may cause problems in the entire integrated energy system, it is extremely necessary to evaluate the health of the devices.
The prognosis and health management is widely applied to an important scientific research field in a new energy system, and can calculate the degradation state of equipment or a system and estimate the residual service life of the system. Thereby improving the safety and troubleshooting rate of the equipment. For example, in a complex system of rotating equipment, prediction of the remaining useful life of the system may help prevent maintenance errors and may also avoid operation under unsafe conditions
In the related art, when evaluating the degradation state of the device, a data set of the device is collected and processed, and then a plurality of processed parameter data are aggregated to form a health index to evaluate the device. However, the health index formed by the above method cannot accurately and comprehensively reflect the state of the equipment, the accuracy of performance degradation state evaluation on the equipment is low, and the prediction precision cannot meet the requirements of actual application scenarios.
Disclosure of Invention
The present application is directed to solving, at least to some extent, one of the technical problems in the related art.
Therefore, a first objective of the present application is to provide an evaluation method for a device performance degradation state based on a combined feature, the method formulates a practical and effective feature parameter evaluation standard based on a feature extraction method of the combined feature, and obtains a feature parameter more suitable for a current scene, so that a current degradation stage of the device can be effectively identified, which is beneficial to realizing maintenance based on a device state, and meanwhile, a prediction result of the remaining service life of the device can be output, thereby facilitating early warning, reducing the maintenance cost of the device, and improving the safety and the troubleshooting rate of the device.
A second objective of the present application is to provide a system for evaluating the performance degradation state of a device based on combined features;
a third object of the present application is to propose a non-transitory computer-readable storage medium.
To achieve the above object, a first aspect of the present application is embodied in a method for evaluating a device performance degradation state based on combined features, the method including the following steps:
acquiring evaluation data of at least one device in a new energy system, wherein the evaluation data comprises device operation data acquired by a data acquisition and monitoring control System (SCADA);
decomposing the evaluation data through Wavelet Packet Decomposition (WPD), and reconstructing corresponding wavelet data for each evaluation data based on the decomposed data;
calculating the energy of each wavelet data, calculating the entropy of each wavelet data according to the energy of each wavelet data, decomposing the entropy of each wavelet data through Empirical Mode Decomposition (EMD), converting the decomposed entropy according to a preset number of inherent modes, and combining each converted entropy to extract a combined feature vector corresponding to the evaluation data;
inputting the combined feature vector to a trained SOM (self-organizing map) and calculating a health evaluation index through the SOM so as to evaluate the performance degradation state of the at least one device through the health evaluation index;
and constructing a residual service life RUL model, and predicting the residual service time of the at least one device according to a health evaluation curve through the residual service life RUL model.
Optionally, in an embodiment of the present application, calculating the health assessment indicator includes: calculating a minimum quantization error MQE based on the combined eigenvector and the weight vector of the best matching unit BMU; and calculating the health assessment index according to the minimum quantization error MQE and the scale parameter of the equipment in the normal state.
Optionally, in an embodiment of the present application, predicting the remaining usage time of the at least one device from the health assessment curve includes: acquiring a health evaluation curve and a preset fault threshold value of equipment; predicting a time at which the health assessment curve reaches the failure threshold to obtain a failure time of the at least one device; subtracting the current time from the time of failure to determine a remaining usage time of the at least one device.
Optionally, in an embodiment of the present application, after constructing the remaining lifetime RUL model, the method further includes: and adding the corresponding residual service life RUL model according to the characteristics of the current evaluation data.
Optionally, in an embodiment of the present application, the evaluation data is decomposed by the following formula:
Figure BDA0003343747820000021
wherein h isk() Representing low-pass filtering, gk() Representing high-pass filtering;
corresponding wavelet data is constructed by the following formula:
Figure BDA0003343747820000031
where h () is a low-pass filter correlation scale function, g () is a high-pass filter correlation scale function, k represents the decomposed part, and i and j are the number of wavelet coefficients.
Optionally, in an embodiment of the present application, the entropy of each of the wavelet data is calculated by the following formula:
Figure BDA0003343747820000032
wherein the content of the first and second substances,
Figure BDA0003343747820000033
where i and j are the number of wavelet coefficients, N is the length of the wavelet data,
Figure BDA0003343747820000034
is the energy of the ith wavelet data of the jth layer,
Figure BDA0003343747820000035
is the probability distribution of the energy of the ith wavelet data of the jth layer.
Optionally, in an embodiment of the present application, the decomposed entropy is converted by the following formula:
Figure BDA0003343747820000036
wherein, ci(t) represents the ith natural mode function at the time t after decomposition, r (t) represents the residual equation at the time t, and L represents the number of preset natural modes.
To achieve the above object, a second aspect of the present application provides an evaluation system for device performance degradation state based on combined features, including the following modules:
the system comprises an acquisition module, a monitoring control system and a monitoring control system, wherein the acquisition module is used for acquiring evaluation data of at least one device in the new energy system, and the evaluation data comprises device operation data acquired by the SCADA (supervisory control and data acquisition) system;
the reconstruction module is used for decomposing the evaluation data through Wavelet Packet Decomposition (WPD) and reconstructing corresponding wavelet data for each evaluation data based on the decomposed data;
the feature extraction module is used for calculating the energy of each wavelet data, calculating the entropy of each wavelet data according to the energy of each wavelet data, decomposing the entropy of each wavelet data through Empirical Mode Decomposition (EMD), converting the decomposed entropy according to a preset number of inherent modes, and combining each converted entropy to extract a combined feature vector corresponding to the evaluation data;
the evaluation module is used for inputting the combined feature vector to a trained SOM (self-organizing map) and calculating a health evaluation index through the SOM so as to evaluate the performance degradation state of the at least one device through the health evaluation index;
and the prediction module is used for constructing a residual service life RUL model, and predicting the residual service time of the at least one device according to the health evaluation curve through the residual service life RUL model.
Optionally, in an embodiment of the present application, the evaluation module is specifically configured to calculate a minimum quantization error MQE according to the combined feature vector and the weight vector of the best matching unit BMU; and calculating the health assessment index according to the minimum quantization error MQE and the scale parameter of the equipment in the normal state.
The technical scheme provided by the embodiment of the application at least has the following beneficial effects: the method comprises the steps of firstly extracting features by utilizing WPD and EMD technologies, extracting corresponding entropy features from a time sequence after wavelet coefficient reconstruction by performing wavelet packet decomposition on an original signal, and extracting an entropy sequence by performing trend analysis through EMD to serve as an input vector of a self-organizing mapping network. And then constructing a health index by utilizing the SOM neural network, wherein the confidence value extracted from the SOM is used for realizing the performance degradation evaluation and the RUL estimation of the equipment and/or the system. Therefore, the evaluation method can effectively identify the current degradation stage of the equipment, accurately reflect the current health state of the equipment, is beneficial to realizing maintenance based on the equipment state, can output the prediction result of the residual service life of the equipment, is convenient for early warning, reduces the maintenance cost of the equipment, and improves the safety and the fault removal rate of the equipment.
In order to implement the foregoing embodiments, the third aspect of the present application further provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method for evaluating the device performance degradation state based on the combined features in the foregoing embodiments is implemented.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The foregoing and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
fig. 1 is a flowchart of a method for evaluating a performance degradation state of a device based on combined features according to an embodiment of the present application;
fig. 2 is a schematic diagram of neural network model training based on joint learning according to an embodiment of the present disclosure;
fig. 3 is a schematic flowchart of a specific method for evaluating a device performance degradation state based on combined features according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of an evaluation system for a device performance degradation state based on combined features according to an embodiment of the present application.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
It should be noted that, in the present application, when the performance degradation state of the device is evaluated in the related art, after the data of a large amount and lacking of corresponding scene experience is processed through the processes of collecting a data set, feature extraction, noise smoothing, feature accumulation, and the like, the generated health index cannot accurately and comprehensively reflect the state of the device, and the accuracy of the performance degradation state evaluation of the device is low, so that a method and a system for evaluating the performance degradation state of the device based on the combined features are provided. According to the method for extracting the features based on the combined features, practical and effective feature parameter evaluation standards are formulated, and feature parameters more suitable for the current scene are obtained, so that the current degradation stage of the equipment can be effectively identified, the maintenance based on the equipment state is facilitated, meanwhile, the prediction result of the residual service life of the equipment can be output, early warning is facilitated, the maintenance cost of the equipment is reduced, and the safety and the troubleshooting rate of the equipment are improved.
The following describes a method and a system for evaluating a device performance degradation state based on combined features according to embodiments of the present invention in detail with reference to the accompanying drawings.
Fig. 1 is a flowchart of a method for evaluating a device performance degradation state based on combined features according to an embodiment of the present application, as shown in fig. 1, the method includes the following steps:
step 101, collecting evaluation data of at least one device in the new energy system, wherein the evaluation data comprises device operation data collected by a data collection and monitoring control system SCADA.
Specifically, acquiring evaluation data of at least one device in the new energy system means that any number of devices in the new energy system can be selected as devices to be evaluated and data of the devices can be acquired according to actual evaluation requirements, and the number of the acquired devices can be one or more of all devices of the new energy system or each device can be subjected to data acquisition and evaluation.
The evaluation data is various data required for calculation when evaluating the performance degradation state of the device, and includes, but is not limited to: equipment operating years, historical maintenance data, current operating data, energy consumption data, and the like. The current operation data refers to data generated in the process flow of the equipment, and may be, for example, the current temperature or pressure of the equipment. The evaluation data of the equipment can be collected in different modes, for example, the operation age data of the equipment is read from a historical database of the equipment, and the like.
In the embodiment of the present application, the current operation Data of the device may be acquired through a Data Acquisition And monitoring Control system (SCADA for short). The SCADA can monitor and control each device in the new energy system on the operation site, and various functions such as data acquisition and device control are achieved. As an example, the SCADA in the present application may be configured as a client/server architecture, wherein the server is connected to preset monitoring devices, and the monitoring devices may be various types of sensors disposed in each operating device and the environment in the new energy system to collect the operating parameters of the devices. The client is used for human-computer interaction, and the evaluation result of the performance degradation state of the equipment can be displayed on a human-computer interaction interface.
Further, errors may occur in the collected evaluation data in practical applications, for example, the collected evaluation data may have phenomena such as missing or abnormal values. Therefore, in order to acquire more accurate evaluation data for subsequent evaluation, in an embodiment of the present application, after the evaluation data is acquired, a corresponding method may be selected for interpolating and removing an abnormal value according to the condition of the currently acquired evaluation data. As another possible implementation manner, for a static device, a regression-type method may be used to perform numerical interpolation and remove abnormal values, for example, a least square or polynomial regression method may be used to perform interpolation on missing data and screen out abnormal values in the data.
Of course, the estimation data may be interpolated and the abnormal value may be removed in other manners, and the specific implementation manner may be determined according to the actual situation, which is not limited herein, for example, the abnormal data of the current device caused by the emergency condition under each working condition may be removed by using the multivariate gaussian mixture model, and the operation data of the device in the normal operation state may be obtained. Thus, by interpolating and removing the abnormal value from the evaluation data, the accuracy of the evaluation of the device performance degradation state can be further improved.
Furthermore, in an embodiment of the present application, in order to improve the efficiency of evaluating the health state of the device, after the evaluation data of the current device is collected, whether the current device fails may be detected first, and if it is determined that the device fails, the operation and maintenance personnel may be directly reminded to report and repair the device, so as to remove the failure in time and ensure the normal operation of the device, and if it is determined that the device does not fail at present, the subsequent evaluation of the performance degradation state of the device may be performed, so as to further understand the degradation state of the device. During specific implementation, a fault threshold value can be set in advance, the acquired data is compared with threshold value data, whether the acquired data exceeds the threshold value or not is judged, and if the acquired data exceeds the threshold value, the fault of the equipment parameter is directly judged.
It should be noted that, because the operation data of the device is generated in real time, the application collects the transformed time series data sets in time sequence, and the evaluation can be performed according to the evaluation data at different time.
And 102, decomposing the evaluation data through Wavelet Packet Decomposition (WPD), and reconstructing corresponding wavelet data for each evaluation data based on the decomposed data.
The wavelet packet decomposition (WPD for short) may analyze a detailed portion of the evaluation data of the present application through multi-iteration wavelet transformation, and project the input evaluation data into a wavelet packet basis function space to perform finer signal analysis. According to the method and the device, the characteristics of the evaluation data acquired in the previous step can be extracted through wavelet packet decomposition, and the useful part aiming at the current scene is obtained through decomposition and reconstruction of the input data, so that the accuracy of the generated health evaluation index of the performance degradation state of the evaluation equipment is improved.
In an embodiment of the present application, when decomposing and reconstructing evaluation data by using a wavelet packet decomposition WPD technique, the evaluation data may be decomposed by using the following formula:
Figure BDA0003343747820000061
wherein h isk() Representing low-pass filtering, gk() Representing high-pass filtering, and then constructing corresponding wavelet data by the following formula:
Figure BDA0003343747820000071
where h () is a low-pass filter correlation scale function, g () is a high-pass filter correlation scale function, k represents the decomposed part, and i and j are the number of wavelet coefficients.
Specifically, one of the collected evaluation data is selected as input data, for example,
Figure BDA0003343747820000072
for example, the evaluation data may be arranged first, and one of the evaluation data in a certain layer may be selected
Figure BDA0003343747820000073
To input data, t is the time of the acquired data. In extracting features, first, the input data is decomposed into high-pass filters by the first formula
Figure BDA0003343747820000074
And low-pass filtering
Figure BDA0003343747820000075
Two filter data, then, according to the decomposed data, reconstructing the corresponding wavelet data for the current input data, i.e. reconstructing the ith wavelet of the j-th layer for the input signal by the second formula, wherein, the low-pass filtering related scale function and the high-pass filtering related wavelet function are both known functions, and the reconstructed wavelet also comprises a high-pass part and a low-pass part, wherein, i and j are the number of wavelet coefficients, i.e. represent the reconstructed wavelet
Figure BDA0003343747820000076
Is the ith wavelet of the jth layer.
Step 103, calculating the energy of each wavelet data, calculating the entropy of each wavelet data according to the energy of each wavelet data, decomposing the entropy of each wavelet data through Empirical Mode Decomposition (EMD), converting the decomposed entropy according to a preset number of inherent modes, and combining the converted entropies to extract a combined feature vector corresponding to the evaluation data.
The Empirical Mode Decomposition (EMD) is a trend analysis method, and can extract an entropy sequence, that is, extract corresponding entropy features from a time sequence after wavelet coefficient reconstruction.
In specific implementation, in order to combine reconstructed multiple data signals, after the energy of each wavelet data is calculated according to the relation between the energy of the signal and the entropy, the entropy of each wavelet data is calculated according to the energy of each wavelet data. Specifically, the entropy of each wavelet data is calculated by the following formula:
Figure BDA0003343747820000077
wherein the content of the first and second substances,
Figure BDA0003343747820000078
where i and j are the number of wavelet coefficients, N is the length of the wavelet data,
Figure BDA0003343747820000079
is the energy of the ith wavelet data of the jth layer,
Figure BDA00033437478200000710
is the probability distribution of the energy of the ith wavelet data of the jth layer.
Thus, the degradation change of the device performance can be reflected based on the calculated wavelet packet energy entropy. After the entropy corresponding to all reconstructed wavelet data is calculated by the method, the sequence of the entropy in the whole time period is s after calculation is carried out according to the data corresponding to each momenten(t)。
Furthermore, decomposing the entropy of each wavelet data through Empirical Mode Decomposition (EMD), converting the decomposed entropy according to a preset number of inherent modes, and combining each converted entropy to extract corresponding entropy characteristics from the time series after wavelet coefficient reconstruction, namely extracting a combined characteristic vector T corresponding to the evaluation data.
It should be noted that the empirical mode decomposition EMD can decompose the signal into a remaining part and a different frequency band part, for example, the original signal can be decomposed by the following formula:
Figure BDA0003343747820000081
where x (t) is the original signal, ci(t) is the i-th natural mode function at time t, rnAnd (t) is a residual equation of the n modes corresponding to the t moment. While sen (T) calculated by the above formula for calculating entropy of each wavelet data represents a sequence of entropy of a full time period, in the present application, a post-preset number (e.g., a preset fixed constant L) of time eigenmodes of sen (T) may be taken to convert sen (T), and each converted entropy is combined as a new combined feature vector T, for example, on the basis of decomposition by the above formula, sen (T) may be converted by the following formula:
Figure BDA0003343747820000082
wherein L represents the number of the preset natural modes, i.e. L natural modes are taken from sen (t), and r (t) represents the residual equation at time t.
Further, after the entropies are sequentially converted by the above method, the converted entropies are combined by the following formula to obtain a combined feature vector T:
Figure BDA0003343747820000083
therefore, according to the evaluation method for the performance degradation state of the equipment, the characteristics of the collected evaluation data are extracted through WPD and EMD technologies, the part which is useful for state evaluation in the input data is obtained through wavelet decomposition of the original signal, corresponding entropy characteristics are extracted from the time sequence after wavelet coefficient reconstruction, and the entropy sequence can be extracted through trend analysis of EMD, so that the characteristic extraction method based on WP-EMD is combined with the traditional root mean square value, kurtosis and peak factor, the degradation characteristics of wavelet packet energy entropy are reflected successfully, the parameters are fused after data reconstruction is achieved, the health index can be established according to the WPD theory subsequently, and a health index system is generated.
And 104, inputting the combined feature vector to a trained SOM (self-organizing map) network, and calculating a health evaluation index through the SOM network so as to evaluate the performance degradation state of at least one device through the health evaluation index.
The SOM can enable neurons in the SOM to recombine themselves, convert an input signal mode of any dimensionality into one-dimensional or two-dimensional discrete mapping, and output a classification result. The combined feature vector T generated in step 103 is used as an input vector of the SOM network in the present application, and a Confidence Value (CV) extracted from the SOM is used as a health assessment index to assess the health state of the equipment.
Specifically, the SOM may be trained in advance according to historical data in an offline stage, or may be trained according to acquired real-time data, and the specific SOM training mode may refer to a mode of training a self-organizing map network in the related art. As one possible implementation manner, the weight of each node may be initialized, a combined feature vector T is randomly selected from a divided training set and presented to the network, and then each node is checked to calculate which node has the weight most similar to the input vector T, that is, the euclidean distance between the input vector T and each weight is calculated, and the neuron with the weight vector most similar to the input is used as an obtained Best Matching Unit (Best Matching Unit, abbreviated as BMU).
Further, after the SOM is trained through the correlation training step, a weight vector mmbu of the best matching unit is determined, where the mmbu is a weight matrix obtained through training and classification of the SOM neural network. In the embodiment of the present application, the input of the SOM neural network is T, and the output is a classification result, which may be divided into two types in the embodiment of the present application, for example, the first m T are one type, and the other T are one type, the specific classification manner may be determined according to the operation site of the system or the requirements of the equipment, and the bmu is the vector of the shortest distance between the determined weight vector and the input T. That is to say, the evaluation method of the present application evaluates the weight vector in the trained SOM model acquisition model, and the specific model training mode is not limited here.
Further, the combined feature vector is input to a trained self-organizing map network SOM, and the health assessment index is calculated through the self-organizing map network SOM. In one embodiment of the present application, calculating the health assessment index may include calculating a minimum quantization error MQE according to the combined eigenvector and the weight vector of the best matching unit BMU, and calculating the health assessment index according to the minimum quantization error MQE and the scale parameter of the device in the normal state.
Specifically, the minimum quantization error may be first calculated by the following formula:
MQE＝||T-mBMU||，
where MQE is the minimum quantization error (minimum quantization error), which is a vector based on time T for quantizing the degradation, bmu is the weight vector of the best matching unit obtained above, and T is the combined feature vector generated in step 103. Then, the confidence value CV of the SOM output is calculated by the following formula:
Figure BDA0003343747820000091
wherein c is a proportional parameter of the device in a normal state, which is a preset fixed constant, and the specific obtaining mode may be experiment on the device to be evaluated on site or obtaining according to historical experience.
Therefore, after the data T is input into the SOM neural network model, a confidence value CV is extracted from the SOM through the SOM neural network training and is used as a health evaluation index to reflect the health state of the equipment, and the current degradation stage of the equipment can be effectively identified through the CV. For example, the corresponding relationship between the health assessment index and the health status of the equipment in different levels may be pre-established, then the section to which the numerical value of the health assessment index of the current equipment belongs is determined, the corresponding health status level of the equipment is determined, and then the performance degradation status of the equipment is analyzed according to the current health status of the equipment. For another example, a performance scoring standard of the equipment may be set according to the health assessment index, a performance score of the current equipment is calculated according to the currently determined health assessment index, the score of the best performance of the equipment in the initial state is compared with the current performance score, and the current performance degradation state of the equipment is determined according to the difference.
And 105, constructing a residual service life RUL model, and predicting the residual service time of at least one device according to the health evaluation curve through the residual service life RUL model.
The Remaining Useful Life (RUL) model is a model for predicting the expected normal operating time of the device from the current moment to the occurrence of the potential fault, and various types of neural network models can be selected according to actual requirements, such as an inverse neural network (BP) or a Fuzzy Neural Network (FNN) and the like, and the model can be modeled as the RUL model. In the embodiment of the application, the health assessment index CV is a vector based on time t, and the application can acquire the transformed time series data sets in time sequence, so that a health assessment curve can be generated according to CV values at different times, and the remaining service time of the equipment corresponding to the input health assessment curve is predicted through the remaining service life RUL model.
In specific implementation, as a possible implementation manner, the remaining service time of the device is predicted according to the health evaluation curve, the health evaluation curve and a preset fault threshold of the device may be obtained first, then the time when the health evaluation curve reaches the fault threshold is predicted to obtain the fault time of the device, and then the current time is subtracted from the fault time to determine the remaining service time of the device. That is to say, in the embodiment of the present application, according to the regressed CV curve, the fault threshold of the device corresponding to the curve may be set, when the CV curve will touch the fault threshold in the future is predicted, and then the possible normal use time of the device is predicted, and then the predicted normal use time is subtracted from the current time to obtain the remaining service life time of the device.
Therefore, the evaluation method can evaluate and set the current performance degradation state, can predict the time when the future equipment fails, and sends the predicted result to operation and maintenance personnel, thereby being beneficial to realizing maintenance based on the equipment state, and accurately and comprehensively describing the degradation evolution process of the equipment by combining the current performance degradation state and the residual service life time of the equipment.
It should be noted that, in practical applications, many types of devices may be evaluated, the types of the evaluation data of different devices are different, and a single RUL model may not be suitable for prediction of multiple types of devices, and therefore, in one embodiment of the present application, after constructing the remaining lifetime RUL model, the corresponding remaining lifetime RUL model may be further added according to the characteristics of the current evaluation data, that is, the RUL model constructed in the present application is not limited to a single model, and can be flexibly configured according to actual needs, for example, the corresponding residual service life prediction algorithm can be determined according to the type of the currently input data, a new prediction model is added, and predicting the current equipment through the corresponding RUL model so as to improve the accuracy and the applicability of predicting the residual service life of the equipment in an actual operation scene.
It should also be noted that, in one embodiment of the present application, the neural network model of the present application may be trained by means of joint learning. Specifically, in the modeling of the conventional prediction model, data required by model training is generally collected into a data center, the model is retrained, and then prediction is performed according to the trained model. In the embodiment of the present application, the model is subjected to horizontal joint learning, that is, the model is trained in a manner that can be regarded as sample-based distributed model training. In specific implementation, as shown in fig. 2, the present application first distributes all data to different machines, each machine downloads a model from a server, trains the model using local data, returns to the server for parameters that need to be updated, and then the server aggregates the returned parameters on each machine, updates the model, and feeds back the updated model to each machine. The specific implementation steps can include: each participant in fig. 2 downloads the latest model from server a; each participant trains a model by using local data, encrypts gradient and uploads the gradient to a server A, and the server A aggregates gradient update model parameters of each user; the server A returns the updated model to each participant, wherein the more the number of the participants is, the more the samples of the model in the server are, and the stronger the adaptability of the returned model is; each participant updates its respective model.
Therefore, the SOM model and the RUL model in the application are established based on the joint learning framework, the adaptability of the models is improved, and the accuracy of evaluation of the performance degradation state of the equipment through each model is further improved.
In summary, in the method for evaluating the device performance degradation state based on the combined features according to the embodiment of the present application, the WPD and the EMD technology are used to extract features, wavelet packet decomposition is performed on an original signal, corresponding entropy features are extracted from a time sequence after wavelet coefficient reconstruction, and trend analysis is performed through the EMD to extract an entropy sequence, which is used as an input vector of the self-organizing mapping network. And then constructing a health index by utilizing the SOM neural network, wherein the confidence value extracted from the SOM is used for realizing the performance degradation evaluation and the RUL estimation of the equipment and/or the system. Therefore, the evaluation method can effectively identify the current degradation stage of the equipment, accurately reflect the current health state of the equipment, is beneficial to realizing maintenance based on the equipment state, can output the prediction result of the residual service life of the equipment, is convenient for early warning, reduces the maintenance cost of the equipment, and improves the safety and the fault removal rate of the equipment.
In order to more clearly illustrate the method for evaluating the performance degradation state of the device based on the combined features according to the embodiment of the present application, a specific evaluation embodiment is described in detail below. Fig. 3 is a flowchart illustrating a specific method for evaluating a device performance degradation state based on a combined feature according to an embodiment of the present disclosure.
As shown in fig. 3, when the evaluation is performed by the method, data is collected in the first step, and according to the characteristics of the equipment to be evaluated at present, the operating age, the historical maintenance data, the current operating data and the energy consumption data of the equipment are selected, and a time period in which the equipment is to be evaluated is selected. Wherein, the operation data is collected by scada for each parameter data of the device. And detecting whether phenomena such as missing or abnormal values appear in the collected data, and if so, carrying out numerical interpolation and removing the abnormal values by combining expert experience or regression methods. And then, judging whether the collected data exceeds a threshold value through a preset threshold value, so as to directly judge whether the equipment parameter has a fault, if so, directly reporting for repair, and if not, entering the next step.
And secondly, extracting the combined features. When the combination characteristics are extracted, firstly, the existing signal data are decomposed, the input data are decomposed according to a wavelet packet decomposition method, and the ith wavelet of the jth layer is reconstructed according to a WPD algorithm. And introducing the relation between the energy of the signal and the entropy, firstly calculating the energy of the reconstructed wavelet, calculating probability distribution according to the energy, and then calculating the entropy corresponding to each signal according to the probability distribution. Further, the entropy corresponding to all the signals is calculated, and a sequence of the entropy in the whole time period is obtained. Since the sequence of the full-time entropy is calculated before, the later L time inherent modes in the sequence of the full-time entropy are taken as a new combined feature vector T through empirical mode decomposition.
And thirdly, modeling the health assessment index. In the second step, the input original signal can be decomposed and recombined to obtain a new signal vector, and a health assessment index model is established in the step, specifically according to the following formula
MQE＝||T-mBMU||
Where MQE is the minimum quantization error used to quantize the degraded case, T is the input data calculated in the second step, and bmu is the weight vector of the best matching unit of T. Further, the CV represents the performance value of the device, and can be calculated by the following formula:
Figure BDA0003343747820000121
where c is the normal condition proportionality parameter obtained from MQE, which is a constant that can be obtained experimentally or empirically by the field device. The CV is obtained by training a SOM neural network, and the input data is T, that is, the calculated performance index or health index of the equipment or some component in the equipment.
And fourthly, carrying out RUL modeling. The RUL modeling has various alternatives, for example, BP (inverse neural network), FNN (fuzzy neural network), etc., can be used as the RUL model. Setting a fault threshold of the equipment according to the regressed CV curve, predicting when the CV curve will reach the fault threshold in the future, and predicting the possible service time of the equipment/component, thereby obtaining the remaining service life time.
And fifthly, outputting the remaining service life result. Inputting the original data of the model trained in the fourth step, processing the data in the second step and the data in the third step, training and testing the model in the fourth step to finally predict the equipment fault at the future moment, subtracting the current time from the predicted time to obtain the result of the residual service life of the equipment/component, and finally outputting the result.
In order to implement the foregoing embodiments, the present application further provides an evaluation system for a device performance degradation state based on a combined feature, and fig. 4 is a schematic structural diagram of the evaluation system for a device performance degradation state based on a combined feature according to an embodiment of the present application, as shown in fig. 4, the system includes an acquisition module 100, a reconstruction module 200, a feature extraction module 300, an evaluation module 400, and a prediction module 500.
The acquisition module 100 is configured to acquire evaluation data of at least one device in the new energy system, where the evaluation data includes device operation data acquired by a SCADA (supervisory control and data acquisition) system.
And the reconstruction module 200 is configured to decompose the evaluation data by wavelet packet decomposition WPD, and reconstruct corresponding wavelet data for each evaluation data based on the decomposed data.
The feature extraction module 300 is configured to calculate energy of each wavelet data, calculate an entropy of each wavelet data according to the energy of each wavelet data, decompose the entropy of each wavelet data through empirical mode decomposition EMD, convert the decomposed entropy according to a preset number of inherent modes, and combine each converted entropy to extract a combined feature vector corresponding to the evaluation data.
And the evaluation module 400 is used for inputting the combined feature vector to the trained SOM, and calculating a health evaluation index through the SOM so as to evaluate the performance degradation state of at least one device through the health evaluation index.
And the prediction module 500 is used for constructing a residual service life RUL model, and predicting the residual service time of at least one device according to the health evaluation curve through the residual service life RUL model.
Optionally, in an embodiment of the present application, the evaluation module is specifically configured to: calculating a minimum quantization error MQE based on the combined eigenvector and the weight vector of the best matching unit BMU; the health assessment indicator is calculated based on the minimum quantization error MQE and the scale parameter for the device in the normal state.
Optionally, in an embodiment of the present application, the prediction module 500 is specifically configured to: acquiring a health evaluation curve and a preset fault threshold value of equipment; predicting a time for the health assessment curve to reach a fault threshold to obtain a fault time for the at least one device; the remaining usage time of the at least one device is determined by subtracting the current time from the time of failure.
Optionally, in an embodiment of the present application, the prediction module 500 is further configured to: and adding the corresponding residual service life RUL model according to the characteristics of the current evaluation data. .
Optionally, in an embodiment of the present application, the reconstruction module 200 is specifically configured to decompose the evaluation data by the following formula:
Figure BDA0003343747820000131
wherein h isk() Representing low-pass filtering, gk() Representing high-pass filtering;
corresponding wavelet data is constructed by the following formula:
Figure BDA0003343747820000132
where h () is a low-pass filter correlation scale function, g () is a high-pass filter correlation scale function, k represents the decomposed part, and i and j are the number of wavelet coefficients.
Optionally, in an embodiment of the present application, the feature extraction module 300 is specifically configured to calculate the entropy of each wavelet data by the following formula:
Figure BDA0003343747820000141
wherein the content of the first and second substances,
Figure BDA0003343747820000142
where i and j are the number of wavelet coefficients, N is the length of the wavelet data,
Figure BDA0003343747820000143
is the energy of the ith wavelet data of the jth layer,
Figure BDA0003343747820000144
is the probability distribution of the energy of the ith wavelet data of the jth layer.
Optionally, in an embodiment of the present application, the feature extraction module 300 is further configured to convert the decomposed entropy by the following formula:
Figure BDA0003343747820000145
wherein, ci(t) represents the ith natural mode function at the time t after decomposition, r (t) represents the residual equation at the time t, and L represents the number of preset natural modes.
It should be noted that the foregoing explanation of the embodiment of the method for evaluating the performance degradation state of a device based on combined features is also applicable to the system of the embodiment, and details are not repeated here
In summary, in the system for evaluating the device performance degradation state based on the combined features according to the embodiment of the present application, the WPD and EMD technologies are first used to extract features, wavelet packet decomposition is performed on an original signal, corresponding entropy features are extracted from a time sequence after wavelet coefficient reconstruction, and trend analysis is performed through EMD to extract an entropy sequence, which is used as an input vector of the self-organizing mapping network. And then constructing a health index by utilizing the SOM neural network, wherein the confidence value extracted from the SOM is used for realizing the performance degradation evaluation and the RUL estimation of the equipment and/or the system. Therefore, practical and effective characteristic parameter evaluation standards are formulated, the degradation change of the equipment is accurately reflected through the wavelet packet energy entropy, the characteristic parameters more suitable for the current scene can be obtained, the current degradation stage of the equipment can be effectively identified, the current health state of the equipment is accurately reflected, the maintenance based on the equipment state is facilitated, meanwhile, the prediction result of the residual service life of the equipment can be output, early warning is facilitated, the maintenance cost of the equipment is reduced, and the safety and the troubleshooting rate of the equipment are improved.
In order to achieve the above embodiments, the present application also proposes a non-transitory computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method for evaluating a device performance degradation state based on combined features as described in any of the above embodiments.
In the description herein, reference to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present application, "plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing steps of a custom logic function or process, and alternate implementations are included within the scope of the preferred embodiment of the present application in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present application.
The logic and/or steps represented in the flowcharts or otherwise described herein, e.g., an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.
It should be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.
It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware related to instructions of a program, which may be stored in a computer readable storage medium, and when the program is executed, the program includes one or a combination of the steps of the method embodiments.
In addition, functional units in the embodiments of the present application may be integrated into one processing module, or each unit may exist alone physically, or two or more units are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. The integrated module, if implemented in the form of a software functional module and sold or used as a stand-alone product, may also be stored in a computer readable storage medium.
The storage medium mentioned above may be a read-only memory, a magnetic or optical disk, etc. Although embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present application, and that variations, modifications, substitutions and alterations may be made to the above embodiments by those of ordinary skill in the art within the scope of the present application.

Claims (10)

1. A method for evaluating the performance degradation state of equipment based on combined characteristics is characterized by comprising the following steps:
acquiring evaluation data of at least one device in a new energy system, wherein the evaluation data comprises device operation data acquired by a data acquisition and monitoring control System (SCADA);
decomposing the evaluation data through Wavelet Packet Decomposition (WPD), and reconstructing corresponding wavelet data for each evaluation data based on the decomposed data;
calculating the energy of each wavelet data, calculating the entropy of each wavelet data according to the energy of each wavelet data, decomposing the entropy of each wavelet data through Empirical Mode Decomposition (EMD), converting the decomposed entropy according to a preset number of inherent modes, and combining each converted entropy to extract a combined feature vector corresponding to the evaluation data;
inputting the combined feature vector to a trained SOM (self-organizing map) and calculating a health evaluation index through the SOM so as to evaluate the performance degradation state of the at least one device through the health evaluation index;
and constructing a residual service life RUL model, and predicting the residual service time of the at least one device according to a health evaluation curve through the residual service life RUL model.
2. The assessment method of claim 1, wherein said calculating a health assessment indicator comprises:
calculating a minimum quantization error MQE based on the combined eigenvector and the weight vector of the best matching unit BMU;
and calculating the health assessment index according to the minimum quantization error MQE and the scale parameter of the equipment in the normal state.
3. The assessment method according to claim 2, wherein said predicting a remaining usage time of said at least one device from a health assessment curve comprises:
acquiring a health evaluation curve and a preset fault threshold value of equipment;
predicting a time at which the health assessment curve reaches the failure threshold to obtain a failure time of the at least one device;
subtracting the current time from the time of failure to determine a remaining usage time of the at least one device.
4. The method of claim 1, wherein after constructing the residual lifetime RUL model, further comprising: and adding the corresponding residual service life RUL model according to the characteristics of the current evaluation data.
5. The evaluation method according to claim 1, wherein the evaluation data is decomposed by the following formula:
Figure FDA0003343747810000021
wherein h isk() Representing low-pass filtering, gk() Representing high-pass filteringWave;
corresponding wavelet data is constructed by the following formula:
Figure FDA0003343747810000022
where h () is a low-pass filter correlation scale function, g () is a high-pass filter correlation scale function, k represents the decomposed part, and i and j are the number of wavelet coefficients.
6. The evaluation method according to claim 1, wherein the entropy of each of the wavelet data is calculated by the following formula:
Figure FDA0003343747810000023
wherein the content of the first and second substances,
Figure FDA0003343747810000024
where i and j are the number of wavelet coefficients, N is the length of the wavelet data,
Figure FDA0003343747810000025
is the energy of the ith wavelet data of the jth layer,
Figure FDA0003343747810000026
is the probability distribution of the energy of the ith wavelet data of the jth layer.
7. The evaluation method according to claim 1, wherein the decomposed entropy is converted by the following formula:
Figure FDA0003343747810000027
wherein, ci(t) represents a result after decompositionthe ith natural mode function at time t, r (t), represents the residual equation at time t, and L represents the number of preset natural modes.
8. A system for evaluating a state of performance degradation of a device based on combined features, comprising:
the system comprises an acquisition module, a monitoring control system and a monitoring control system, wherein the acquisition module is used for acquiring evaluation data of at least one device in the new energy system, and the evaluation data comprises device operation data acquired by the SCADA (supervisory control and data acquisition) system;
the reconstruction module is used for decomposing the evaluation data through Wavelet Packet Decomposition (WPD) and reconstructing corresponding wavelet data for each evaluation data based on the decomposed data;
the feature extraction module is used for calculating the energy of each wavelet data, calculating the entropy of each wavelet data according to the energy of each wavelet data, decomposing the entropy of each wavelet data through Empirical Mode Decomposition (EMD), converting the decomposed entropy according to a preset number of inherent modes, and combining each converted entropy to extract a combined feature vector corresponding to the evaluation data;
the evaluation module is used for inputting the combined feature vector to a trained SOM (self-organizing map) and calculating a health evaluation index through the SOM so as to evaluate the performance degradation state of the at least one device through the health evaluation index;
and the prediction module is used for constructing a residual service life RUL model, and predicting the residual service time of the at least one device according to the health evaluation curve through the residual service life RUL model.
9. The evaluation system of claim 8, wherein the evaluation module is specifically configured to:
calculating a minimum quantization error MQE based on the combined eigenvector and the weight vector of the best matching unit BMU;
and calculating the health assessment index according to the minimum quantization error MQE and the scale parameter of the equipment in the normal state.
10. A non-transitory computer-readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the method for combined feature-based assessment of device performance degradation status according to any of claims 1-7.
CN202111316146.3A 2021-11-08 2021-11-08 Method and system for evaluating equipment performance degradation state based on combination characteristics Pending CN114201825A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111316146.3A CN114201825A (en) 2021-11-08 2021-11-08 Method and system for evaluating equipment performance degradation state based on combination characteristics

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111316146.3A CN114201825A (en) 2021-11-08 2021-11-08 Method and system for evaluating equipment performance degradation state based on combination characteristics

Publications (1)

Publication Number Publication Date
CN114201825A true CN114201825A (en) 2022-03-18

Family

ID=80647118

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111316146.3A Pending CN114201825A (en) 2021-11-08 2021-11-08 Method and system for evaluating equipment performance degradation state based on combination characteristics

Country Status (1)

Country Link
CN (1) CN114201825A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115619106A (en) * 2022-12-19 2023-01-17 中国人民解放军火箭军工程大学 Method and system for determining quantity of spare parts of laser gyroscope in consideration of performance degradation

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115619106A (en) * 2022-12-19 2023-01-17 中国人民解放军火箭军工程大学 Method and system for determining quantity of spare parts of laser gyroscope in consideration of performance degradation
CN115619106B (en) * 2022-12-19 2023-05-16 中国人民解放军火箭军工程大学 Method and system for determining number of spare parts of laser gyroscope in consideration of performance degradation

Similar Documents

Publication Publication Date Title
CN109298697B (en) Method for evaluating working state of each part of thermal power plant system based on dynamic baseline model
Hu et al. Online performance assessment method for a model-based prognostic approach
Lapira et al. Wind turbine performance assessment using multi-regime modeling approach
CN111597682B (en) Method for predicting remaining life of bearing of gearbox of wind turbine
CN110414154B (en) Fan component temperature abnormity detection and alarm method with double measuring points
KR20140041766A (en) Method of sequential kernel regression modeling for forecasting and prognostics
CN110398368B (en) FBM-based bearing inner ring fault residual life prediction method based on long correlation model
CN112416643A (en) Unsupervised anomaly detection method and unsupervised anomaly detection device
CN113036913B (en) Method and device for monitoring state of comprehensive energy equipment
CN114239377A (en) Method and system for evaluating health state of urban rail electromechanical equipment and storage medium
WO2023065580A1 (en) Fault diagnosis method and apparatus for gearbox of wind turbine generator set
CN112416662A (en) Multi-time series data anomaly detection method and device
CN111382494A (en) System and method for detecting anomalies in sensory data of industrial machines
CN113470694A (en) Remote listening monitoring method, device and system for hydraulic turbine set
CN110969185A (en) Equipment abnormal state detection method based on data reconstruction
CN114266301A (en) Intelligent power equipment fault prediction method based on graph convolution neural network
CN114580291A (en) LSTM-VAE-based mobile equipment health condition assessment method
CN114201825A (en) Method and system for evaluating equipment performance degradation state based on combination characteristics
CN115081647A (en) Industrial intelligent instrument fault pre-diagnosis method based on Bayesian network model
CN116821610B (en) Method for optimizing wind power generation efficiency by utilizing big data
CN117290685A (en) Power plant power equipment expert diagnosis system and method based on historical data
KR20210097613A (en) Power consumption prediction system for optimization of demand management resources
CN117371207A (en) Extra-high voltage converter valve state evaluation method, medium and system
CN117077532A (en) Multi-model fusion method for life prediction of wind turbine generator
CN116914917A (en) Big data-based monitoring and management system for operation state of power distribution cabinet

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