Power equipment fault state monitoring method and device, electronic equipment and storage medium

Abstract

The invention provides a power equipment fault state monitoring method and device, electronic equipment and a storage medium, and belongs to the technical field of power system monitoring, and the method comprises the steps: inputting a power feature vector corresponding to each power equipment in a power grid node into a risk prediction model, and obtaining a risk prediction value of the power grid node; if the risk prediction value is smaller than a preset threshold value, monitoring the fault state of the power equipment based on the power feature vector corresponding to each power equipment; and if the risk prediction value is greater than or equal to a preset threshold value, monitoring the fault state of the power equipment based on the power feature vector corresponding to each power equipment and a preset power equipment fault knowledge graph. According to the power equipment fault state monitoring method and device, the electronic equipment and the storage medium provided by the invention, the power equipment fault state monitoring efficiency and accuracy are improved.

Description

Technical Field

[0001] The present disclosure belongs to the technical field of power system monitoring, and more specifically, relates to a method for monitoring the fault status of power equipment. Background Art

[0002] With the large-scale production of electricity, power equipment is becoming increasingly complex, with increasingly sophisticated functions and a higher degree of automation. The interdependencies between subsystems are becoming increasingly close. Failure of any part of the equipment during operation can disrupt production, causing significant economic losses and even catastrophic consequences. Therefore, online monitoring of power equipment is necessary to support condition-based maintenance strategies and intelligent operation and maintenance. Current online monitoring systems for power equipment, which measure a single state parameter, only monitor faulty equipment and ignore healthy equipment. Summary of the Invention

[0003] The purpose of the present disclosure is to provide a method and device for monitoring the fault status of electric equipment, an electronic device, and a storage medium, which improve the efficiency and accuracy of monitoring the fault status of electric equipment.

[0004] A first aspect of the embodiments of the present disclosure provides a method for monitoring a fault state of an electric power device, comprising: Inputting the power feature vector corresponding to each power device in the power grid node into the risk prediction model to obtain the risk prediction value of the power grid node; If the risk prediction value is less than a preset threshold, the fault status of each power device is monitored based on the power characteristic vector corresponding to the power device; If the risk prediction value is greater than or equal to a preset threshold, the fault status of the power equipment is monitored based on the power feature vector corresponding to each power equipment and a preset power equipment fault knowledge graph.

[0005] A second aspect of the embodiments of the present disclosure provides a device for monitoring a fault state of an electric power device, comprising: The risk prediction module is used to input the power feature vector corresponding to each power device in the power grid node into the risk prediction model to obtain the risk prediction value of the power grid node; A first execution module is configured to monitor the fault status of each power device based on the power characteristic vector corresponding to the power device if the risk prediction value is less than a preset threshold; The second execution module is used to monitor the fault status of the power equipment based on the power feature vector corresponding to each power equipment and the preset power equipment fault knowledge graph if the risk prediction value is greater than or equal to the preset threshold.

[0006] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of the above-mentioned method for monitoring the fault status of power equipment when executing the computer program.

[0007] According to a fourth aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned method for monitoring the fault status of electric power equipment are implemented.

[0008] The beneficial effects of the power equipment fault status monitoring method and device, electronic device, and storage medium provided by the embodiments of the present disclosure are: by inputting the power characteristic vector into the risk prediction model, the monitoring mode can be switched according to the risk prediction value of the power grid node; when the risk prediction value is less than the preset threshold, direct monitoring based on the power characteristic vector can efficiently screen out equipment with a high probability of normal operation, saving monitoring resources. When the risk prediction value is greater than or equal to the preset threshold, the power equipment fault knowledge graph is introduced, and the topological relationship, fault mode and other information clearly depicted by it are used to comprehensively consider the equipment fault status. This method can not only accurately judge the fault, but also understand the fault propagation path and impact range through the fault propagation impact chain, which helps to provide early warning and avoid the spread of faults. The present disclosure implements layered processing for different risk levels, which improves the accuracy and efficiency of monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0010] Figure 1 A flowchart of a method for monitoring a fault state of an electric power device according to an embodiment of the present disclosure is provided; Figure 2 A structural block diagram of a device for monitoring fault status of electric power equipment according to an embodiment of the present disclosure; Figure 3 A schematic block diagram of an electronic device provided in one embodiment of the present disclosure. DETAILED DESCRIPTION

[0011] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present disclosure. However, it will be apparent to those skilled in the art that the present disclosure may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present disclosure with unnecessary detail.

[0012] In order to make the purpose, technical solutions and advantages of the present disclosure more clear, specific embodiments will be described below with reference to the accompanying drawings.

[0013] Please refer to Figure 1 , Figure 1 A flow chart of a method for monitoring a fault state of an electric power device according to an embodiment of the present disclosure is provided, wherein the method comprises: S101: Inputting the power feature vector corresponding to each power device in the power grid node into the risk prediction model to obtain the risk prediction value of the power grid node.

[0014] In this embodiment, various sensors are used to collect real-time operating parameter data for each electrical device in a power grid node. This operating parameter data includes current, voltage, temperature, vibration, and partial discharge. This embodiment first preprocesses the collected data to obtain a power feature vector containing multi-dimensional information such as voltage, current, power factor, and temperature. Preprocessing includes data cleaning and normalization. During the data cleaning process, this embodiment can identify and eliminate abnormal data caused by sensor failure, communication interference, and other reasons, such as current and voltage values that clearly exceed the normal range. The preprocessed operating data is then fused and normalized to obtain a power feature vector. The fusion process integrates the different electrical parameters according to a fusion method or strategy, enabling them to collaborate with each other to form an information set that more comprehensively and accurately reflects the status. The fusion process in this embodiment can employ methods such as weighted averaging, Kalman filtering, and neural network fusion.

[0015] The present invention can input the power feature vector into the risk prediction model to obtain the risk prediction value of the power grid node, or derive evaluation indicators such as the equipment's operating status failure probability and equipment health; the risk prediction model can be constructed based on a machine learning algorithm (such as a decision tree, support vector machine, neural network, etc.) or a deep learning algorithm (such as a long short-term memory network), and the model's training data comes from the historical operating data and failure case data of the power equipment.

[0016] S102: If the risk prediction value is less than a preset threshold, the fault status of each power device is monitored based on the power feature vector corresponding to the power device.

[0017] In this embodiment, if the resulting risk prediction value is less than a preset threshold value, determined based on a combination of factors such as the grid's long-term stable operation experience, historical equipment failure data, and industry standards, it indicates that the current grid node is relatively stable. Based on the power feature vectors corresponding to each power device, a feature-based analysis algorithm designed specifically for fault monitoring of that device is used to conduct detailed monitoring of the device's fault status.

[0018] In this embodiment, the methods for determining the preset threshold include: setting the threshold based on statistical historical data, setting the threshold based on model performance, setting the threshold by combining expert experience and dynamic feedback, setting the threshold based on multi-objective optimization threshold decision, etc.

[0019] Among them, setting the threshold based on statistical historical data includes: collecting risk prediction values of power grid nodes in historical operation (data under normal and fault conditions); statistically distributing risk values under normal conditions (such as taking the 95% quantile as the threshold) to ensure that the probability of failure is extremely low when it is below the threshold; combined with fault sample verification, if the threshold is too low, it will lead to missed reports, and if it is too high, it will frequently trigger the enhanced mode.

[0020] In this embodiment, setting the threshold based on model performance includes: using the validation set data to calculate the risk prediction value of the model output and marking the real fault label; drawing the ROC curve (receiver operating characteristic curve), selecting the point with the largest Youden index (i.e., the maximum difference between the true positive rate and the false positive rate) or setting an acceptable false alarm rate (such as 5%) according to business needs, and reversely deriving the corresponding threshold.

[0021] In this embodiment, the threshold is set based on a threshold decision based on multi-objective optimization, including: modeling the threshold selection as an optimization problem while considering constraints such as computing resources, fault detection rate, and response time; defining the objective function: maximizing the fault detection rate and minimizing the computing overhead; introducing constraints: the computing resource upper limit of the enhanced mode and the fault response time requirement; and using genetic algorithms or particle swarm optimization to solve the optimal threshold.

[0022] S103: If the risk prediction value is greater than or equal to a preset threshold, the fault status of the power equipment is monitored based on the power feature vector corresponding to each power equipment and a preset power equipment fault knowledge graph.

[0023] In this embodiment, if the risk prediction value is greater than or equal to a preset threshold, it indicates a high risk for the power grid node, requiring more comprehensive and in-depth monitoring. In this case, a knowledge graph inference algorithm is used to comprehensively monitor the fault status of each power device based on the power feature vector corresponding to each device and a preset power device fault knowledge graph. The power device fault knowledge graph includes rich information such as various types of power device faults, fault causes, fault symptoms, and topological relationships between devices.

[0024] From the above, it can be concluded that the method for monitoring the fault status of power equipment provided by the embodiment of the present disclosure can switch the monitoring mode according to the risk prediction value of the power grid node by inputting the power feature vector into the risk prediction model; when the risk prediction value is less than the preset threshold, direct monitoring based on the power feature vector can efficiently screen out equipment with a high probability of normal operation, saving monitoring resources. When the risk prediction value is greater than or equal to the preset threshold, the power equipment fault knowledge graph is introduced, and the topological relationship, fault mode and other information clearly depicted by it are used to comprehensively consider the equipment fault status. This method can not only accurately judge the fault, but also understand the fault propagation path and impact range through the fault propagation impact chain, which helps to provide early warning and avoid the spread of faults. At the same time, the entire method processes different risk levels in a layered manner to improve the accuracy and efficiency of monitoring, provide strong guarantees for the stable operation of the power system, and effectively reduce the economic losses and power supply interruption risks caused by equipment failures.

[0025] In one embodiment of the present disclosure, fault status monitoring of each power device based on the power characteristic vector corresponding to the power device includes: Inputting the power feature vector of the target power equipment into a preset fault classification model to obtain a fault probability value of the target power equipment, where the target power equipment is a power equipment in a power grid node; If the fault probability value is greater than or equal to a preset fault probability threshold, the target power equipment is monitored for a fault state based on the power monitoring data of the target power equipment and the fault probability value; If the fault probability value is less than the preset fault probability threshold, the target power equipment is monitored for fault status based on the matching degree between the power feature vector of the target power equipment and the preset feature vector library; the preset feature vector library includes the power feature vector of the target power equipment in a non-fault state.

[0026] In this embodiment, a target power device in a power grid node is selected and its power feature vector is input into a fault classification model. The model identifies, analyzes, and calculates the features of the input vector, outputting a quantitative fault probability value that reflects the target power device's fault state. The fault classification model is a machine learning classification algorithm (e.g., a random forest classifier) trained using power device fault data and normal operation data.

[0027] In this embodiment, if the fault probability value is greater than or equal to the fault probability threshold, it indicates a high risk of equipment failure. At this point, real-time, high-precision power monitoring data is collected from the target power equipment, including detailed information such as real-time current fluctuations and voltage transients during equipment operation. Combined with the fault probability value obtained above, a fault diagnosis algorithm, such as a Bayesian network-based fault diagnosis method, is used to monitor the fault status of the target power equipment, accurately locate the fault type and degree, or increase the frequency of detection of the target power equipment. Based on the fault type and degree, the type and location of the fault are provided to power equipment maintenance personnel, thereby ensuring the normal operation of the power equipment.

[0028] In this embodiment, the distribution pattern of the corresponding eigenvectors when the fault occurs is statistically analyzed through a database of historical equipment fault cases, and the 95% quantile of the fault occurrence probability is taken as the initial threshold. Gaussian kernel density estimation (KDE) is performed on the eigenvectors of the normal operating state of the equipment, and the probability density function of the non-fault state is calculated and compared with the probability density function of the fault state. The intersection of the two (i.e., the Bayesian minimum error rate point) is taken as the fault probability threshold.

[0029] In this embodiment, the fault probability threshold is dynamically optimized based on the machine learning model. When training the fault classification model, the optimal threshold is determined by the receiver operating characteristic (ROC) curve and the optimal balance point: the true positive rate and false positive rate under different thresholds are calculated; the threshold that maximizes the Youden index (true positive rate - false positive rate) is selected as the fault probability threshold; In this embodiment, a time series sliding window (e.g., a 30-day window) is used to calculate the mean and standard deviation of the equipment failure probability within the window in real time. The calculation formula for dynamically setting the failure probability threshold is:

[0030] in, is the failure probability threshold, k is the adjustment coefficient, is the mean of the probability values within the window, is the standard deviation of the probability values within the window.

[0031] Set the threshold for critical equipment to a low threshold (e.g., 0.7); set the threshold for non-critical equipment to a high threshold (e.g., 0.9); adjust the threshold based on the equipment aging model and the predicted remaining useful life. The calculation formula is:

[0032] in, is the failure probability threshold, is the baseline threshold of the nth device, is the remaining useful life of the nth device, is the design life of the nth device.

[0033] In this embodiment, if the fault probability value is less than the preset fault probability threshold, it means that the equipment is currently in normal operation, but further investigation of potential hidden dangers is still required. At this time, the power feature vector of the target power equipment is compared with a pre-constructed preset feature vector library containing power feature vectors of the target power equipment in a non-fault state under different working conditions. By calculating the Euclidean distance, cosine similarity and other indicators between the vectors, the matching degree between the power feature vector of the target power equipment and the preset feature vector library is obtained. According to the matching degree value, the target power equipment is monitored for fault status using a monitoring algorithm based on pattern recognition, which will promptly detect any abnormal change trends, generate alarm information or increase the frequency of sensor monitoring of the power equipment. This embodiment can also introduce an equipment operating life correction factor when calculating the matching degree, and relax the matching degree threshold for old equipment by 15%-20%.

[0034] In one embodiment of the present disclosure, inputting the power feature vector of the target power equipment into a preset fault classification model to obtain the fault probability value of the target power equipment includes: fusing the environmental data around the target power equipment and the feature vector corresponding to the target power equipment to obtain target data; The target data is input into the preset fault classification model to obtain the fault probability value of the target power equipment.

[0035] In this embodiment, environmental monitoring sensors are used to collect real-time environmental data around the target power equipment, including information such as ambient temperature, humidity, and electromagnetic interference intensity. Feature encoding is performed on the meteorological data (temperature, humidity, wind speed) and electromagnetic environment indicators (field strength, interference spectrum) around the equipment to generate an environmental feature vector. In this embodiment, a data fusion algorithm, such as a Kalman filter-based data fusion method, is used to organically fuse the environmental data surrounding the target power equipment with the feature vector corresponding to the target power equipment to generate target data that includes the equipment's own status and external environmental influencing factors. A weighted multimodal fusion method based on an attention mechanism: The weight ratio of the power feature vector to the environmental feature vector is dynamically adjusted through a gated fusion unit, where the weight of the environmental feature is increased by 30%-50% in high temperature and high humidity environments. In this embodiment, before the target data is input into the fault classification model, a generative adversarial network is used to expand minority class fault samples to generate first data. This first data is then input into a pre-built, pre-trained fault classification model that has undergone multiple rounds of optimization training. This model, based on a deep learning neural network architecture, such as a convolutional neural network or a recurrent neural network, extracts and analyzes complex features in the first data layer by layer, ultimately outputting a fault probability value that accurately reflects the fault state of the target power equipment.

[0036] In this embodiment, a regression model between thresholds and environmental conditions is established using equipment operating environment parameters (temperature, humidity, load rate) as covariates:

[0037] in is the failure probability threshold, is the benchmark threshold, α is the environmental impact coefficient, β is the load impact coefficient, is the environmental parameter, is the load current.

[0038] In one embodiment of the present disclosure, fault status monitoring of each power device is performed based on the power feature vector corresponding to each power device and a preset power device fault knowledge graph, including: Match the power feature vector of the target power equipment with the fault mode node in the power equipment fault knowledge graph to obtain a target fault mode set, where the target power equipment is the power equipment in the power grid node; According to the target fault mode set and the fault propagation impact chain, the fault status of the target power equipment is obtained; The fault propagation impact chain is constructed based on the topological relationship of power equipment in the power equipment fault knowledge graph. It uses the breadth-first search algorithm to traverse potential fault propagation paths and combines it with the dynamic Bayesian network to calculate the fault impact weight of each fault propagation path.

[0039] In this embodiment, a detailed feature vector is obtained for the target power equipment in the power grid node. The vector contains key elements such as the electrical characteristics, operating parameters, and structural information of the equipment. At the same time, a pre-built power equipment fault knowledge graph stored in the form of a graph database is called. A knowledge graph matching algorithm, such as a matching algorithm based on semantic similarity, is used to match the power feature vector of the target power equipment with the fault mode nodes in the power equipment fault knowledge graph one by one. During the matching process, each dimension of information in the equipment feature vector is compared and analyzed with the fault characteristics represented by the fault mode node. After complex calculations, a target fault mode set with a high correlation with the target power equipment feature vector is obtained.

[0040] The method for constructing the power equipment fault knowledge graph is as follows: constructing a triple (equipment-failure mode-influencing factor) based on the equipment ontology library, where the failure mode node includes: fault feature vector template, typical fault waveform spectrum and maintenance case library index; An improved weighted directed graph model is used to generate the fault propagation impact chain. The topological correlation strength is calculated based on the electrical distance between devices, and the propagation path weight is determined by combining the cross-correlation analysis of the fault recording data. When updating the dynamic Bayesian network, the real-time load change rate is introduced as a conditional variable. When the regional load mutation exceeds 10%, the propagation path weight is dynamically recalculated. The fault status assessment of the target power equipment adopts an evidence reasoning mechanism: the matching target fault mode set is probabilistically synthesized with the parent node fault evidence on the propagation chain, and when there is a conflict in the cross-device fault evidence, the conflict resolution algorithm based on information entropy is activated.

[0041] In this embodiment, the fault mode set is a set of candidate fault modes that match the characteristics of the target power equipment, filtered from the power equipment fault knowledge graph. It covers the types of faults that can occur in the power system, their characteristic parameters, and associated information. Fault types include physical types (such as short circuit, overload, insulation breakdown, ground fault, and harmonic distortion) and subcategories (for example, short circuit types include three-phase short circuit, single-phase ground fault, and phase-to-phase short circuit; overload types include instantaneous overload, sustained overload, and thermal overload; and insulation faults include partial discharge, surface flashover, and internal moisture).

[0042] The empty association information includes: marking the location level of the fault in the power grid topology (such as transmission line, substation bus, distribution network feeder end) and equipment type (transformer, circuit breaker, capacitor bank, etc.).

[0043] The temporal laws describing fault development include: progressive faults caused by insulation aging (resistance value decreases exponentially); transient faults caused by lightning strikes; and cumulative faults caused by overload (temperature rise curve conforms to the exponential model).

[0044] The relationship between fault propagation and impact includes: the propagation path weight is based on the fault propagation probability calculated by the dynamic Bayesian network, such as the downstream bus voltage drop due to transformer overload and the backup protection over-tripping due to circuit breaker refusal to operate.

[0045] In this embodiment, based on the topological relationship of the power equipment clearly depicted in the power equipment fault knowledge graph, a breadth-first search algorithm is used to comprehensively traverse potential fault propagation paths starting from the target fault mode set. At the same time, by using the dynamic Bayesian network and the time series factors in the equipment operation process and the uncertainty in the fault propagation process, the fault impact weight of each fault propagation path is calculated through learning and analysis of historical fault data. Combining the target fault mode set with the constructed fault propagation impact chain, a fault diagnosis algorithm based on knowledge reasoning is used, such as an algorithm based on a combination of rule reasoning and case reasoning, to finally accurately obtain the fault status of the target power equipment, including the fault type, fault location, and the impact range of the fault.

[0046] In this embodiment, based on the fault modes in the target fault mode set, the fault propagation paths and their fault impact weights associated with them are searched in the fault propagation impact chain. By comprehensively considering these factors, the probability that the target power device is in a fault state can be calculated. For example, if the path corresponding to a certain fault mode in the fault propagation impact chain has a high fault impact weight, and this fault mode is identified in the target fault mode set, then it can be inferred that the target power device is more likely to have failed. Conversely, if the fault impact weight of the associated path is low, or if the fault modes in the target fault mode set do not have a strong correlation in the fault propagation impact chain, then the probability that the target power device has failed is relatively low. Ultimately, based on these calculation and analysis results, the fault state of the target power device is determined, and it is judged whether it is in a fault state and the severity of the fault.

[0047] In this embodiment, the construction of the fault propagation impact chain includes: power grid topology modeling, potential path search, propagation weight calculation and dynamic probability modeling; based on the real-time data of the Common Information Model standard, a topology graph containing the following elements is constructed: nodes, edges and attributes; dynamic weights are assigned to the edges to obtain power grid topology modeling, wherein the dynamic weights assigned to the edges reflect the coupling strength between devices.

[0048] Specifically, nodes include: generators, transformers, busbars, circuit breakers and other equipment; edges include: transmission lines, relay protection logic relationships, and electrical coupling relationships; attributes include: equipment rated parameters (such as capacity, impedance) and real-time status (load rate, temperature).

[0049] Table 1 Search methods corresponding to different path types

[0050] In this embodiment, as shown in Table 1, which shows the search methods corresponding to different path types, the present invention uses an improved breadth-first search algorithm (BFS with Adaptive Pruning) for multimodal path detection. For electrical connection path detection, the fault source node is used as the starting point, and the neighborhood is expanded along the topological edges. For implicit coupling path identification, the resonant correlation between devices is calculated using the frequency domain impedance matrix.

[0051] In one embodiment of the present disclosure, the power feature vector of the target power equipment is matched with the fault mode nodes in the power equipment fault knowledge graph to obtain a target fault mode set, including: Calculating multiple target similarities between the power feature vector of the target power equipment and the similarity between the fault mode nodes in the power equipment fault knowledge graph; Multiple target similarities are sorted, and fault mode nodes whose target similarities are greater than a first similarity threshold are selected to generate a target fault mode set.

[0052] In this embodiment, similarity calculation algorithms, such as the cosine similarity algorithm and the Pearson correlation coefficient algorithm, are used to calculate the similarity between the target power equipment's power feature vector and each failure mode node in the power equipment failure knowledge graph. During the calculation process, the characteristic information contained in the feature vector and the failure mode node is standardized to ensure the accuracy and comparability of the calculation results.

[0053] The calculated multiple target similarities are sorted in descending order. The first similarity threshold is determined based on statistical analysis of historical equipment failure data and combined with the accuracy requirements for equipment failure monitoring. Fault mode nodes with target similarities greater than the first similarity threshold are selected and aggregated to generate a target fault mode set that accurately reflects the potential failure modes of the target power equipment.

[0054] In one embodiment of the present disclosure, matching the power feature vector of the target power equipment with the fault mode nodes in the power equipment fault knowledge graph to obtain a target fault mode set further includes: adjusting a first similarity threshold according to a load rate of the target power equipment; When the load rate of the target power equipment is greater than or equal to the first load rate, lowering the first similarity threshold according to the first step; When the load rate of the target electrical equipment is less than the first load rate, the first similarity threshold is increased according to the first step.

[0055] In this embodiment, load rate data of the target power equipment is collected in real time. This data is obtained by calculating the real-time operating power and rated power of the equipment and can intuitively reflect the workload of the equipment. The first similarity threshold is dynamically adjusted according to the load rate of the target power equipment.

[0056] When the load rate of the target power equipment is greater than or equal to a first load rate pre-set based on the equipment's rated parameters, historical operating data, and equipment reliability requirements, it indicates that the equipment is operating at high load. At this point, the likelihood of equipment failure increases. To improve the sensitivity of fault monitoring, a subtraction operation is used to lower the first similarity threshold, based on a pre-set first step that balances monitoring accuracy and false alarm rate. This allows more fault mode nodes with a certain degree of correlation with the equipment's feature vector to be included in the target fault mode set, thereby more comprehensively troubleshooting potential fault hazards.

[0057] When the load rate of the target power equipment is less than the first load rate, the equipment is operating at a relatively low load, and the probability of failure is relatively low. To avoid wasted resources and false alarms caused by excessive monitoring, the first similarity threshold is increased using an addition operation based on the first step. This ensures that only failure mode nodes with a high degree of similarity to the equipment's feature vector are selected for inclusion in the target failure mode set, ensuring accurate and effective monitoring.

[0058] In one embodiment of the present disclosure, the method for monitoring a fault state of an electric power device further includes: Generate a feature vector corresponding to each power device according to multiple monitoring data in the power grid node, where the multiple monitoring data are data monitored by multiple sensors on each power device; Perform feature extraction on each monitoring data to obtain the feature vector corresponding to each monitoring data; Vector concatenation is performed on feature vectors corresponding to a plurality of monitoring data of the electric power equipment to obtain an electric power feature vector corresponding to the electric power equipment.

[0059] In this embodiment, multiple types of sensors, such as current sensors, voltage sensors, and temperature sensors, are deployed in different locations within power grid nodes to conduct comprehensive, real-time monitoring of each power device and obtain multiple monitoring data. This data covers various aspects of the device's operation, including electrical parameters, physical state parameters, and environmental parameters.

[0060] For each piece of monitoring data, we apply feature extraction algorithms, such as those based on Fourier transforms in the frequency domain or wavelet transforms in the time-frequency domain, to extract feature vectors from the raw monitoring data that effectively represent the equipment's operating status. During the extraction process, we select appropriate feature extraction methods based on the characteristics of different types of monitoring data to ensure that the extracted feature vectors are representative and effective.

[0061] The feature vectors corresponding to multiple monitoring data points for power equipment are arranged in an orderly manner. Using a vector concatenation algorithm, such as direct concatenation or weighted fusion-based concatenation, these feature vectors are concatenated according to specific rules. This yields a single power feature vector corresponding to the power equipment that comprehensively reflects its operating status and incorporates the characteristics of multi-source monitoring data. This power feature vector serves as an important data foundation for subsequent fault monitoring and analysis.

[0062] The power equipment fault status monitoring method provided by the disclosed embodiments uses a risk prediction model to assess the risk level of power grid nodes in real time, enabling dynamic switching of monitoring strategies. Combining a knowledge graph of power equipment faults with feature vector analysis, a dual-engine monitoring system is constructed, breaking through the limitations of a single model and significantly improving the accuracy and comprehensiveness of fault diagnosis in high-risk scenarios. It can also identify potential fault areas in advance, providing accurate decision-making support for operations and maintenance personnel, quickly locating the root cause of the fault and generating targeted maintenance recommendations, thereby reducing power outage time and lowering operation and maintenance costs.

[0063] Corresponding to the method for monitoring the fault status of electric power equipment in the above embodiment, Figure 2 This is a structural block diagram of a device for monitoring the fault status of an electric power device according to an embodiment of the present disclosure. For ease of explanation, only the parts related to the embodiment of the present disclosure are shown. Figure 2 The power equipment fault status monitoring device 20 includes: a risk prediction module 21, a first execution module 22 and a second execution module 23.

[0064] The risk prediction module 21 is used to input the power feature vector corresponding to each power device in the power grid node into the risk prediction model to obtain the risk prediction value of the power grid node; A first execution module 22 is configured to monitor the fault status of each power device based on the power feature vector corresponding to the power device if the risk prediction value is less than a preset threshold; The second execution module 23 is used to monitor the fault status of the power equipment based on the power feature vector corresponding to each power equipment and the preset power equipment fault knowledge graph if the risk prediction value is greater than or equal to the preset threshold.

[0065] In one embodiment of the present disclosure, the first execution module 22 is specifically configured to: Inputting the power feature vector of the target power equipment into a preset fault classification model to obtain a fault probability value of the target power equipment, where the target power equipment is a power equipment in a power grid node; If the fault probability value is greater than or equal to a preset fault probability threshold, the target power equipment is monitored for a fault state based on the power monitoring data of the target power equipment and the fault probability value; If the fault probability value is less than the preset fault probability threshold, the target power equipment is monitored for fault status based on the matching degree between the power feature vector of the target power equipment and the preset feature vector library; the preset feature vector library includes the power feature vector of the target power equipment in a non-fault state.

[0066] In one embodiment of the present disclosure, the first execution module 22 is specifically configured to: fusing the environmental data around the target power equipment and the feature vector corresponding to the target power equipment to obtain target data; The target data is input into the preset fault classification model to obtain the fault probability value of the target power equipment.

[0067] In one embodiment of the present disclosure, the second execution module 23 is specifically configured to: Match the power feature vector of the target power equipment with the fault mode node in the power equipment fault knowledge graph to obtain a target fault mode set, where the target power equipment is the power equipment in the power grid node; According to the target fault mode set and the fault propagation impact chain, the fault status of the target power equipment is obtained; The fault propagation impact chain is constructed based on the topological relationship of power equipment in the power equipment fault knowledge graph. It uses the breadth-first search algorithm to traverse potential fault propagation paths and combines it with the dynamic Bayesian network to calculate the fault impact weight of each fault propagation path.

[0068] In one embodiment of the present disclosure, the second execution module 23 is specifically configured to: Calculating multiple target similarities between the power feature vector of the target power equipment and the fault mode nodes in the power equipment fault knowledge graph; Multiple target similarities are sorted, and fault mode nodes whose target similarities are greater than a first similarity threshold are selected to generate a target fault mode set.

[0069] In one embodiment of the present disclosure, the first execution module 23 is specifically configured to: adjusting the first similarity threshold according to the load rate of the target power equipment; When the load rate of the target power equipment is greater than or equal to the first load rate, lowering the first similarity threshold according to the first step; When the load rate of the target electrical equipment is less than the first load rate, the first similarity threshold is increased according to the first step.

[0070] In one embodiment of the present disclosure, the risk prediction module 21 is specifically configured to: Generate a feature vector corresponding to each power device according to multiple monitoring data in the power grid node, where the multiple monitoring data are data monitored by multiple sensors on each power device; Perform feature extraction on each monitoring data to obtain the feature vector corresponding to each monitoring data; Vector concatenation is performed on feature vectors corresponding to a plurality of monitoring data of the electric power equipment to obtain an electric power feature vector corresponding to the electric power equipment.

[0071] In this embodiment, by inputting the power characteristic vector into the risk prediction model, the monitoring mode can be switched according to the risk prediction value of the power grid node; when the risk prediction value is less than the preset threshold, direct monitoring based on the power characteristic vector can efficiently screen out equipment with a high probability of normal operation, saving monitoring resources. When the risk prediction value is greater than or equal to the preset threshold, the power equipment fault knowledge graph is introduced, and the topological relationship, fault mode and other information clearly depicted are used to comprehensively consider the equipment fault status. The present invention can not only accurately judge the fault, but also understand the fault propagation path and impact range through the fault propagation impact chain, which helps to provide early warning and avoid the spread of faults. At the same time, the entire method processes different risk levels in a layered manner, improves the accuracy and efficiency of monitoring, provides a strong guarantee for the stable operation of the power system, and effectively reduces the economic losses and power supply interruption risks caused by equipment failures.

[0072] See also Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided by an embodiment of the present disclosure. Figure 3 The electronic device 300 in the embodiment shown may include: one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memory 304 is used to store computer programs, which include program instructions. The processor 301 is used to execute the program instructions stored in the memory 304. The processor 301 is configured to call the program instructions to execute the functions of the modules in the above-mentioned device embodiments, such as Figure 2 The functions of the risk prediction module 21, the first execution module 22 and the second execution module 23 are shown.

[0073] It should be understood that in the embodiments of the present disclosure, the processor 301 may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0074] The input device 302 may include a touchpad, a fingerprint collection sensor (for collecting user fingerprint information and fingerprint direction information), a microphone, etc. The output device 303 may include a display (LCD, etc.), a speaker, etc.

[0075] The memory 304 may include a read-only memory and a random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include a non-volatile random access memory. For example, the memory 304 may also store device type information.

[0076] In a specific implementation, the processor 301, input device 302, and output device 303 described in the embodiments of the present disclosure can execute the implementation methods described in the first and second embodiments of the method for monitoring the fault status of electric power equipment provided in the embodiments of the present disclosure, and can also execute the implementation methods of the electronic device described in the embodiments of the present disclosure, which will not be repeated here.

[0077] In another embodiment of the present disclosure, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program. The computer program includes program instructions. When the program instructions are executed by a processor, all or part of the process of the method in the above embodiment is implemented. The computer program can also be used to instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of each of the above method embodiments are implemented. The computer program includes computer program code, which can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium.

[0078] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the aforementioned embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card, etc. Furthermore, the computer-readable storage medium can include both an internal storage unit of the electronic device and an external storage device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or is about to be output.

[0079] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the composition and steps of each example according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this disclosure.

[0080] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the electronic devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0081] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces or units, or can be an electrical, mechanical or other form of connection.

[0082] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, i.e., they may be located in one place or distributed across multiple network units. Some or all of these units may be selected based on actual needs to achieve the objectives of the embodiments of the present disclosure.

[0083] In addition, the functional units in the various embodiments of the present disclosure may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0084] The above are only specific embodiments of the present disclosure, but the scope of protection of the present disclosure is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or replacements within the technical scope disclosed in this disclosure, and such modifications or replacements should be included in the scope of protection of the present disclosure. Therefore, the scope of protection of the present disclosure should be based on the scope of protection of the claims.

Claims

1. A method for monitoring the fault status of power equipment, characterized in that: include: Inputting the power feature vector corresponding to each power device in the power grid node into the risk prediction model to obtain the risk prediction value of the power grid node; If the risk prediction value is less than a preset threshold, the fault status of each power device is monitored based on the power characteristic vector corresponding to the power device; If the risk prediction value is greater than or equal to a preset threshold, the fault status of the power equipment is monitored based on the power feature vector corresponding to each power equipment and a preset power equipment fault knowledge graph.

2. The method for monitoring the fault status of electric power equipment according to claim 1, wherein: The performing fault status monitoring on each power device based on the power characteristic vector corresponding to the power device includes: Inputting a power feature vector of a target power device into a preset fault classification model to obtain a fault probability value of the target power device, wherein the target power device is a power device in the power grid node; If the fault probability value is greater than or equal to a preset fault probability threshold, performing fault status monitoring on the target power equipment based on the power monitoring data of the target power equipment and the fault probability value; If the fault probability value is less than a preset fault probability threshold, the target power equipment is monitored for fault status based on the matching degree between the power feature vector of the target power equipment and a preset feature vector library; the preset feature vector library includes the power feature vector of the target power equipment in a non-fault state.

3. The method for monitoring the fault status of electric power equipment according to claim 2, wherein: Inputting the power feature vector of the target power equipment into a preset fault classification model to obtain a fault probability value of the target power equipment includes: fusing environmental data around the target power equipment and a power feature vector corresponding to the target power equipment to obtain target data; The target data is input into a preset fault classification model to obtain a fault probability value of the target power equipment.

4. The method for monitoring the fault status of electric power equipment according to claim 1, wherein: The fault status monitoring of each power device based on the power feature vector corresponding to each power device and the preset power device fault knowledge graph includes: Matching a power feature vector of a target power device with a fault mode node in the power device fault knowledge graph to obtain a target fault mode set, wherein the target power device is a power device in the power grid node; Obtaining a fault state of the target power equipment according to the target fault mode set and the fault propagation impact chain; The fault propagation impact chain is constructed based on the topological relationship of the power equipment in the power equipment fault knowledge graph, using a breadth-first search algorithm to traverse potential fault propagation paths, and combining a dynamic Bayesian network to calculate the fault impact weight of each fault propagation path.

5. The method for monitoring the fault status of electric power equipment according to claim 4, wherein: The matching of the power feature vector of the target power equipment with the fault mode nodes in the power equipment fault knowledge graph to obtain a target fault mode set includes: Calculating multiple target similarities between the power feature vector of the target power equipment and the fault mode nodes in the power equipment fault knowledge graph; Multiple target similarities are sorted, and fault mode nodes whose target similarities are greater than a first similarity threshold are selected to generate a target fault mode set.

6. The method for monitoring the fault status of electric power equipment according to claim 5, wherein: Also includes: adjusting the first similarity threshold according to the load rate of the target power equipment; When the load rate of the target power equipment is greater than or equal to the first load rate, lowering the first similarity threshold according to the first step; When the load rate of the target electrical equipment is less than the first load rate, the first similarity threshold is increased according to the first step.

7. The method for monitoring the fault status of electric power equipment according to claim 2, wherein: Also includes: Generate a power feature vector corresponding to each power device according to multiple monitoring data in the power grid node, wherein the multiple monitoring data are data monitored by multiple sensors on each power device; Perform feature extraction on each monitoring data to obtain the feature vector corresponding to each monitoring data; Vector concatenation is performed on feature vectors corresponding to a plurality of monitoring data of the electric power equipment to obtain an electric power feature vector corresponding to the electric power equipment.

8. A device for monitoring the fault status of electric power equipment, characterized in that: include: The risk prediction module is used to input the power feature vector corresponding to each power device in the power grid node into the risk prediction model to obtain the risk prediction value of the power grid node; A first execution module is configured to monitor the fault status of each power device based on the power characteristic vector corresponding to the power device if the risk prediction value is less than a preset threshold; The second execution module is used to monitor the fault status of the power equipment based on the power feature vector corresponding to each power equipment and the preset power equipment fault knowledge graph if the risk prediction value is greater than or equal to the preset threshold.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images