Fault identification method and system for a relay protection device, device, medium

Abstract

The present application relates to relay detection technical field, specifically, it is fault identification method and system, equipment, medium for relay protection equipment, adopt the method provided by the present application, mainly including after the pre-processing of operating parameter and image parameter, data diagnosis model and image diagnosis model are constructed respectively, first fault index is outputted based on operating parameter through data diagnosis model, second fault index is outputted based on image parameter through the image diagnosis model;Fusion fault index is obtained based on first fault index and second fault index, and first judgment threshold and second judgment threshold are set, through the above scheme, the scheme can no longer rely on the device itself alarm information, but objectively evaluates various operating parameters and appearance, to obtain more accurate fault identification result, avoids the best maintenance period.

Description

Technical Field

[0001] This invention relates to the field of relay testing technology, and more specifically, to a fault identification method, system, device, and medium for relay protection equipment. Background Technology

[0002] Relay protection equipment is a core secondary device that ensures the safe and stable operation of the power grid. Aging of the equipment, damage to the circuit boards, and wiring defects can easily lead to protection failure or maloperation, causing power grid accidents.

[0003] Faults are mainly classified into hardware faults, software and program faults, secondary circuit faults, and latent anomalies. Current technologies rely on the alarm signals of relay equipment itself and do not perform real-time analysis of the currently collected data, which may lead to misjudgment and delay the best maintenance time. Furthermore, because fault diagnosis relies mainly on device self-testing and manual inspection, it can only identify obvious alarm faults and is not capable of identifying latent problems such as sampling drift, terminal burning, and capacitor bulging. Summary of the Invention

[0004] The purpose of this invention is to provide a fault identification method, system, device, and medium for relay protection equipment to solve the above-mentioned problems in the prior art.

[0005] This invention is achieved through the following technical solution: In a first aspect, the present invention provides a fault identification method for relay protection equipment, comprising: Obtain the current operating parameters and image parameters of the relay protection device, determine whether a fault signal from the relay protection device has been received, and if a fault information is received, send an alarm signal to the remote control terminal. If no fault signal is received, after preprocessing the operating parameters and image parameters, a data diagnostic model and an image diagnostic model are constructed respectively. Based on the operating parameters, the data diagnostic model outputs the first fault index, and based on the image parameters, the image diagnostic model outputs the second fault index. A fusion fault index is obtained based on the first fault index and the second fault index. A first judgment threshold and a second judgment threshold are set. If the fusion fault index is greater than the first judgment threshold, an alarm signal is sent to the remote control terminal. If the fusion index is less than or equal to the first judgment threshold but greater than the second judgment threshold, the current data acquisition interval is corrected. If the fusion index is less than or equal to the second judgment threshold, the current state is maintained and operating parameters and image parameters are continuously acquired.

[0006] Preferably, the step of outputting the first fault index based on the data diagnostic model using operating parameters includes: Construct an LSTM unit and a deviation-aware loss function, reconstruct the input unit based on the LSTM unit output, calculate the window-length moving average error, and calculate the first fault index based on the reconstructed input unit and the window-length moving average error.

[0007] Preferably, the LSTM unit includes:

[0008]

[0009] In the formula, This represents the cell state after time t. For the Gate of Oblivion This represents the cell state at time t-1. For input gate, Candidate cell state, For time-series residual memory coefficients, In hidden state, For output gate; Based on the hidden state output, a temporal hidden representation is obtained through the encoder layer, and the temporal hidden representation is decoded by the decoder to obtain the reconstructed input unit.

[0010] Preferably, the bias-perceiving loss function includes:

[0011]

[0012]

[0013] In the formula, For the total loss function, Based on the reconstruction loss, This is a penalty item for orientation deviation. and To balance hyperparameters, It is a set of temporal hidden representations. This represents the total time step of the time series samples. For the actual input parameters at time t, To reconstruct the input unit, This is the fault direction mask vector. For the actual set of input parameters, The set of input cells to be reconstructed.

[0014] Preferably, the calculation of the first fault index based on the moving average error of the reconstructed input unit and window length includes:

[0015]

[0016]

[0017] In the formula, To reconstruct the error vector at a single time step, The moving average error is the window length. For window length, To sum and iterate through the indices, The first failure index, These are the baseline parameters for normal operating conditions. Steepness coefficient, .

[0018] Preferably, the step of outputting the second fault index based on image parameters through the image diagnostic model includes: The image parameters are divided into M. For M, non-overlapping windows are used. Window attention weights are set, and self-attention is calculated for each window to output enhanced multi-scale features. The enhanced multi-scale features of several windows and the global downsampling features are fused to obtain the fused features; The second fault index is obtained by mapping the temperature coefficient to sigmoid based on the fusion features.

[0019] In the formula, The second failure factor, It is the sigmoid activation function. For learnable weight vectors, For bias terms, For temperature hyperparameters, .

[0020] Preferably, the correction of the current data collection interval includes:

[0021]

[0022] In the formula, For integration index, and To calculate the weights, their sum is 1. This is the corrected data collection interval duration. This is the initial data collection interval. This is the second judgment threshold.

[0023] Secondly, the present invention also provides a fault identification system for relay protection equipment, for performing the above-described fault identification method for relay protection equipment, comprising: The data processing module is configured to acquire the current operating parameters and image parameters of the relay protection device, determine whether a fault signal of the relay protection device has been received, and if a fault signal is received, send an alarm signal to the remote control terminal; if no fault signal is received, after preprocessing the operating parameters and image parameters, a data diagnostic model and an image diagnostic model are constructed respectively. Based on the operating parameters, the data diagnostic model outputs a first fault index, and based on the image parameters, the image diagnostic model outputs a second fault index. The judgment module is configured to obtain a fused fault index based on a first fault index and a second fault index, and to set a first judgment threshold and a second judgment threshold. If the fused fault index is greater than the first judgment threshold, an alarm signal is sent to the remote control terminal. If the fused index is less than or equal to the first judgment threshold but greater than the second judgment threshold, the current data acquisition interval is corrected. If the fused index is less than or equal to the second judgment threshold, the current state is maintained and the operating parameters and image parameters are continuously acquired.

[0024] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described fault identification method for relay protection equipment.

[0025] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described fault identification method for relay protection equipment.

[0026] The technical solution of the present invention has at least the following advantages and beneficial effects: The method provided by this invention mainly includes preprocessing operating parameters and image parameters, constructing a data diagnostic model and an image diagnostic model respectively, outputting a first fault index based on the operating parameters through the data diagnostic model, and outputting a second fault index based on the image parameters through the image diagnostic model; obtaining a fused fault index based on the first fault index and the second fault index, and setting a first judgment threshold and a second judgment threshold. Through the above scheme, this scheme can no longer rely solely on the alarm information of the equipment itself, but objectively evaluate various operating parameters and appearance to obtain more accurate fault identification results and avoid delaying the best maintenance period. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0030] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. The naming or numbering of steps in this application does not imply that the steps in the method flow must be executed in the chronological / logical order indicated by the naming or numbering. The execution order of named or numbered process steps can be changed according to the desired technical objective, as long as the same or similar technical effect is achieved.

[0031] Furthermore, the connection, coupling, or communication in this application can be a direct connection, coupling, or communication between related objects, or an indirect connection, coupling, or communication through other devices. Moreover, the connection, coupling, or communication between objects can be electrical or other similar forms, and this application does not impose any limitations on these. Independently described modules or sub-modules may or may not be physically separated; they may be implemented in software or hardware, and some modules or sub-modules may be implemented in software, with the processor calling the software to implement the function of these modules or sub-modules, while other modules or sub-modules may be implemented in hardware, such as through hardware circuits. Furthermore, some or all of the modules can be selected to achieve the purpose of this application's solution according to actual needs.

[0032] Please refer to Figures 1-2 This invention provides a fault identification method for relay protection equipment, comprising: S1: Obtain the current operating parameters and image parameters of the relay protection device, determine whether a fault signal of the relay protection device has been received, and if a fault information is received, send an alarm signal to the remote control terminal. In this embodiment, based on the substation data acquisition terminal and high-definition industrial camera equipment, multi-dimensional operating parameters and appearance image parameters of the relay protection equipment are acquired simultaneously. The operating parameters include core timing data such as the equipment's working power supply voltage, three-phase sampling current and voltage, board operating temperature, communication error rate, time synchronization deviation, and switch input and output status. The image parameters include high-definition real-scene images of key parts such as the equipment panel, wiring terminals, capacitor boards, and secondary circuit interfaces.

[0033] While acquiring data, the system monitors in real time fault signals such as built-in self-test messages from the relay protection equipment, background remote signaling alarms, and fault changes, enabling real-time monitoring of visible faults. If the system successfully receives valid fault information from the equipment, including various fault messages such as hardware fault alarms, sampling anomalies, communication interruptions, and circuit faults, it can immediately trigger an emergency alarm mechanism without performing subsequent complex model diagnostic calculations. The system encapsulates key information such as fault type, fault occurrence time, equipment number, and fault location into an alarm signal and uploads it to the remote control terminal in real time. This allows maintenance personnel to detect visible faults immediately and conduct emergency repairs, effectively preventing power grid safety accidents caused by protection failures or maloperations. This step prioritizes responding to hard faults, significantly shortening the alarm response time for visible faults and ensuring the fault tolerance and safety of the power grid operation.

[0034] S2: If no fault signal is received, after preprocessing the operating parameters and image parameters, a data diagnostic model and an image diagnostic model are constructed respectively. Based on the operating parameters, the first fault index is output through the data diagnostic model, and based on the image parameters, the second fault index is output through the image diagnostic model. Specifically, if the system does not detect any explicit fault signals from the relay protection equipment, it indicates that the equipment does not have any sudden hard faults, but there is a potential risk of latent faults and gradual defects. At this time, the dual-dimensional data processing and intelligent diagnosis process is initiated.

[0035] First, the collected raw operating parameters and image parameters are standardized and preprocessed. For time-series operating parameters, missing value filling, outlier removal, normalization scaling, and time-series smoothing filtering are used to eliminate data noise caused by substation electromagnetic interference and data acquisition equipment jitter, ensuring the continuity and accuracy of time-series data. For equipment appearance image parameters, image noise reduction, size standardization, grayscale correction, and region of interest cropping are performed to remove redundant background information and focus on key defect areas such as board ablation, capacitor bulging, and terminal blackening.

[0036] After preprocessing, the pre-trained and optimized data diagnostic model and image diagnostic model are used to conduct independent diagnostic analyses. The data diagnostic model takes time-series operating parameters as input, mines implicit features such as gradual drift of equipment parameters and abnormal operating states, and optimizes the calculation through a directional bias loss function to output a first fault index representing the risk of equipment operating status. The image diagnostic model takes the preprocessed equipment appearance image as input, and accurately identifies minor appearance defects through multi-scale local and global feature fusion and window attention enhancement mechanisms, outputting a second fault index representing the risk of physical appearance failure of the equipment. The two models are independent and complementary in dimensions, avoiding the problems of missed detections by single data diagnostics and false detections by single image diagnostics, and providing accurate two-dimensional fault quantification basis for subsequent fusion analysis.

[0037] S3: Based on the first fault index and the second fault index, a fusion fault index is obtained, and a first judgment threshold and a second judgment threshold are set. If the fusion fault index is greater than the first judgment threshold, an alarm signal is sent to the remote control terminal. If the fusion index is less than or equal to the first judgment threshold but greater than the second judgment threshold, the current data acquisition interval is corrected. If the fusion index is less than or equal to the second judgment threshold, the current state is maintained, and operating parameters and image parameters are continuously acquired.

[0038] Based on the first and second fault indices output by the dual models, the system achieves integrated fault risk assessment and adaptive adjustment of equipment data acquisition strategies, constructing a hierarchical fault early warning and routine monitoring mechanism. The system integrates the two-dimensional fault indices through a weighted fusion algorithm, sets adaptive weights based on the operating characteristics of relay protection equipment, and calculates a fused fault index that comprehensively reflects the overall operating status of the equipment, thus achieving a comprehensive quantitative assessment of fault risk.

[0039] Simultaneously, two levels of quantitative judgment thresholds are preset: a first judgment threshold for high-risk alarms and a second judgment threshold for risk warnings, forming a three-level status discrimination mechanism. When the fusion fault index exceeds the first judgment threshold, the equipment is judged to have a high probability of hidden faults or potential operational defects, which can easily trigger abnormal protection functions. The system immediately generates detailed alarm information and uploads it to the remote control terminal, reminding maintenance personnel to carry out special inspections and defect investigations.

[0040] When the fusion fault index falls between the second and first judgment thresholds, the equipment is determined to be in a sub-healthy operating state, exhibiting low-risk hazards such as minor parameter drift and slight external wear. The system adaptively shortens the data and image acquisition interval, increasing the frequency of equipment status monitoring and enabling dynamic tracking of potential hazards. When the fusion fault index is less than or equal to the second judgment threshold, the equipment is determined to be in good operating condition with no potential faults. The system maintains the original acquisition interval and monitoring mode, continuously and stably acquiring equipment operation and image data, achieving routine, intelligent, and differentiated status monitoring and fault assessment of relay protection equipment.

[0041] The method provided by this invention mainly includes preprocessing operating parameters and image parameters, constructing a data diagnostic model and an image diagnostic model respectively, outputting a first fault index based on the operating parameters through the data diagnostic model, and outputting a second fault index based on the image parameters through the image diagnostic model; obtaining a fused fault index based on the first fault index and the second fault index, and setting a first judgment threshold and a second judgment threshold. Through the above scheme, this scheme can no longer rely solely on the alarm information of the equipment itself, but objectively evaluate various operating parameters and appearance to obtain more accurate fault identification results and avoid delaying the best maintenance period.

[0042] In one exemplary embodiment of the present invention, the step of outputting a first fault index based on operating parameters through a data diagnostic model includes: Construct an LSTM unit and a deviation-aware loss function, reconstruct the input unit based on the LSTM unit output, calculate the window-length moving average error, and calculate the first fault index based on the reconstructed input unit and the window-length moving average error.

[0043] Specifically, the LSTM unit includes:

[0044]

[0045] In the formula, This represents the cell state after time t. For the Gate of Oblivion This represents the cell state at time t-1. For input gate, Candidate cell state, For time-series residual memory coefficients, In hidden state, For output gate; Based on the hidden state output, a temporal hidden representation is obtained through the encoder layer, and the temporal hidden representation is decoded by the decoder to obtain the reconstructed input unit.

[0046] Bias-aware loss functions include:

[0047]

[0048]

[0049] In the formula, For the total loss function, Based on the reconstruction loss, This is a penalty item for orientation deviation. and To balance hyperparameters, It is a set of temporal hidden representations. This represents the total time step of the time series samples. For the actual input parameters at time t, To reconstruct the input unit, This is the fault direction mask vector. For the actual set of input parameters, The set of input cells to be reconstructed.

[0050] The calculation of the first fault index for the reconstructed input cell and window length moving average error includes:

[0051]

[0052]

[0053] In the formula, To reconstruct the error vector at a single time step, The moving average error is the window length. For window length, To sum and iterate through the indices, The first failure index, These are the baseline parameters for normal operating conditions. Steepness coefficient, .

[0054] The second fault index output by the image diagnostic model based on image parameters includes: The image parameters are divided into M. For M, non-overlapping windows are used. Window attention weights are set, and self-attention is calculated for each window to output enhanced multi-scale features. The enhanced multi-scale features of several windows and the global downsampling features are fused to obtain the fused features; The second fault index is obtained by mapping the temperature coefficient to sigmoid based on the fusion features.

[0055] In the formula, The second failure factor, It is the sigmoid activation function. For learnable weight vectors, For bias terms, For temperature hyperparameters, .

[0056] In an exemplary embodiment of the present invention, the correction of the current data collection interval duration includes:

[0057]

[0058] In the formula, For integration index, and To calculate the weights, their sum is 1. This is the corrected data collection interval duration. This is the initial data collection interval. This is the second judgment threshold.

[0059] The above model uses a floating and varying data acquisition interval to cope with changes in the current device status. A specific adjustment cycle can be set according to the specific usage, and the current data acquisition interval can be fixed within an adjustment cycle.

[0060] When the equipment is operating well, a larger acquisition interval can reduce the frequency of acquiring operating parameters and appearance images, thereby reducing the workload of the acquisition terminal and camera device, reducing the generation of massive amounts of redundant data, saving network transmission bandwidth and backend database storage resources, and reducing long-term system power consumption and maintenance costs. When the equipment is in a sub-healthy state and the fusion fault index is between two threshold levels, the acquisition interval is automatically shortened to increase the data and image sampling density, intensively capturing parameter drift and the slow deterioration process of appearance defects, and obtaining more time series and image samples. This facilitates the dual diagnostic model to continuously track the development trend of potential problems, capture slowly changing and intermittent hidden faults, and reduce the probability of missed detection.

[0061] The present invention also provides a fault identification system for relay protection equipment, for performing the above-described fault identification method for relay protection equipment, comprising: The data processing module is configured to acquire the current operating parameters and image parameters of the relay protection device, determine whether a fault signal of the relay protection device has been received, and if a fault signal is received, send an alarm signal to the remote control terminal; if no fault signal is received, after preprocessing the operating parameters and image parameters, a data diagnostic model and an image diagnostic model are constructed respectively. Based on the operating parameters, the data diagnostic model outputs a first fault index, and based on the image parameters, the image diagnostic model outputs a second fault index. The judgment module is configured to obtain a fused fault index based on a first fault index and a second fault index, and to set a first judgment threshold and a second judgment threshold. If the fused fault index is greater than the first judgment threshold, an alarm signal is sent to the remote control terminal. If the fused index is less than or equal to the first judgment threshold but greater than the second judgment threshold, the current data acquisition interval is corrected. If the fused index is less than or equal to the second judgment threshold, the current state is maintained and the operating parameters and image parameters are continuously acquired.

[0062] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0063] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. This computer software product, stored in a storage medium, includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0064] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A fault identification method for relay protection equipment, characterized in that, include: Obtain the current operating parameters and image parameters of the relay protection device, determine whether a fault signal from the relay protection device has been received, and if a fault information is received, send an alarm signal to the remote control terminal. If no fault signal is received, after preprocessing the operating parameters and image parameters, a data diagnostic model and an image diagnostic model are constructed respectively. Based on the operating parameters, the first fault index is output through the data diagnostic model, and based on the image parameters, the second fault index is output through the image diagnostic model. A fusion fault index is obtained based on the first fault index and the second fault index. A first judgment threshold and a second judgment threshold are set. If the fusion fault index is greater than the first judgment threshold, an alarm signal is sent to the remote control terminal. If the fusion index is less than or equal to the first judgment threshold but greater than the second judgment threshold, the current data acquisition interval is corrected. If the fusion index is less than or equal to the second judgment threshold, the current state is maintained and operating parameters and image parameters are continuously acquired.

2. The fault identification method for relay protection equipment according to claim 1, characterized in that, The first fault index output by the data diagnostic model based on operating parameters includes: Construct an LSTM unit and a deviation-aware loss function, reconstruct the input unit based on the LSTM unit output, calculate the window-length moving average error, and calculate the first fault index based on the reconstructed input unit and the window-length moving average error.

3. The fault identification method for relay protection equipment according to claim 2, characterized in that, The LSTM unit includes: In the formula, This represents the cell state after time t. For the Gate of Oblivion This represents the cell state at time t-1. For input gate, Candidate cell state, For time-series residual memory coefficients, In hidden state, For output gate; Based on the hidden state output, a temporal hidden representation is obtained through the encoder layer, and the temporal hidden representation is decoded by the decoder to obtain the reconstructed input unit.

4. The fault identification method for relay protection equipment according to claim 3, characterized in that, The deviation-sensing loss function includes: In the formula, For the total loss function, Based on the reconstruction loss, This is a penalty item for orientation deviation. and To balance hyperparameters, It is a set of temporal hidden representations. This represents the total time step of the time series samples. For the actual input parameters at time t, To reconstruct the input unit, This is the fault direction mask vector. For the actual set of input parameters, The set of input cells to be reconstructed.

5. The fault identification method for relay protection equipment according to claim 4, characterized in that, The first fault index is calculated based on the moving average error of the reconstructed input unit and window length, including: In the formula, To reconstruct the error vector at a single time step, The moving average error is the window length. For window length, To sum and iterate through the indices, The first failure index, These are the baseline parameters for normal operating conditions. Steepness coefficient, .

6. The fault identification method for relay protection equipment according to claim 5, characterized in that, The second fault index output by the image diagnostic model based on image parameters includes: The image parameters are divided into M. For M, non-overlapping windows are used. Window attention weights are set, and self-attention is calculated for each window to output enhanced multi-scale features. The enhanced multi-scale features of several windows and the global downsampling features are fused to obtain the fused features; The second fault index is obtained by mapping the temperature coefficient to sigmoid based on the fusion features. In the formula, The second failure factor, It is the sigmoid activation function. For learnable weight vectors, For bias terms, For temperature hyperparameters, .

7. The fault identification method for relay protection equipment according to claim 6, characterized in that, The correction of the current data collection interval duration includes: In the formula, For integration index, and To calculate the weights, their sum is 1. This is the corrected data collection interval. This is the initial data collection interval. This is the second judgment threshold.

8. A fault identification system for relay protection equipment, characterized in that, The method for fault identification of relay protection equipment according to any one of claims 1-7 includes: The data processing module is configured to acquire the current operating parameters and image parameters of the relay protection device, determine whether a fault signal of the relay protection device has been received, and if a fault signal is received, send an alarm signal to the remote control terminal; if no fault signal is received, after preprocessing the operating parameters and image parameters, a data diagnostic model and an image diagnostic model are constructed respectively. Based on the operating parameters, the data diagnostic model outputs a first fault index, and based on the image parameters, the image diagnostic model outputs a second fault index. The judgment module is configured to obtain a fused fault index based on a first fault index and a second fault index, and to set a first judgment threshold and a second judgment threshold. If the fused fault index is greater than the first judgment threshold, an alarm signal is sent to the remote control terminal. If the fused index is less than or equal to the first judgment threshold but greater than the second judgment threshold, the current data acquisition interval is corrected. If the fused index is less than or equal to the second judgment threshold, the current state is maintained and the operating parameters and image parameters are continuously acquired.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the fault identification method for relay protection equipment as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, which, when executed by a processor, implements the fault identification method for relay protection equipment as described in any one of claims 1-7.

Images