Intelligent substation fault identification method and system based on virtual circuit topology

Abstract

The present invention discloses a method and system for intelligent substation fault identification based on virtual circuit topology, relating to the technical field of substation fault identification. The method and system include the following steps: obtaining a first model and a second model, and constructing an identification pattern, wherein the identification pattern includes independent identification and combined identification; identifying substation faults using the identification pattern, and regulating substation equipment nodes based on the identification results; the identification pattern is switched via a control switching system; the control switching system is obtained by obtaining model identification features and constructing it based on a random forest, and the control switching system includes verification of the identification results and system updates. This application is used to solve the problem of balancing the error rate and timeliness caused by dynamic topological changes in traditional single models.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer substation fault identification, and more particularly to a method and system for intelligent transformer substation fault identification based on virtual circuit topology. Background Art

[0002] With the widespread application of smart substations in power systems, their virtual circuits, as digital communication links connecting secondary equipment, undertake key functions such as relay protection and monitoring signals. Traditional fault identification methods mainly rely on reasoning systems based on expert rules to quickly locate faults through preset logical judgments. However, when faced with new fault modes or complex operating conditions, such methods are difficult to adapt to the dynamic changes in the power grid due to problems such as delayed rule base updates and rigid reasoning logic. On the other hand, although deep learning-based recognition models can mine fault characteristics through historical data, they have limitations such as high training data annotation costs, lack of real-time performance, and weak scenario migration capabilities. In particular, in the scenario of multi-source heterogeneous data fusion and analysis in smart substations, the recognition efficiency of a single model is difficult to guarantee.

[0003] The digital transformation of smart substations has made virtual circuit faults dynamic, hidden, and multi-source, further highlighting the limitations of traditional methods. Existing technical solutions generally lack in-depth utilization of real-time circuit topology characteristics. For example, key indicators such as the network redundancy and signal path complexity of virtual circuits are not included in the decision-making basis, resulting in a significant increase in the misjudgment rate in high-redundancy or heterogeneous equipment-intensive scenarios. In addition, most existing systems use static strategies to perform fault identification and are unable to dynamically adjust the combination of identification strategies based on the historical accuracy of the model, the complexity of real-time operating conditions, etc., and lack a closed-loop optimization mechanism based on repair effects. It is difficult to cope with the challenges brought about by the diversification of equipment types and the complexity of communication protocols during substation upgrades and renovations.

[0004] For example, the invention patent publication number CN118818970A discloses a real-time monitoring method and system based on interval observation. This invention provides a real-time monitoring method and system based on interval observation, belonging to the field of intelligent monitoring technology. The method includes the following steps: S1: obtaining the real-time operating parameters of the monitored object and dividing the real-time operating parameters into several intervals based on a preset interval threshold; S2: performing statistical analysis on the real-time operating parameters within each interval to obtain a statistical characteristic value for each interval; S3: comparing the statistical characteristic value for each interval with a preset standard model to calculate the deviation for each interval; S4: determining whether the operating status of the monitored object is abnormal based on the deviation for each interval, and if so, issuing a warning signal; S5: automatically adjusting the control parameters of the monitored object based on the warning signal to restore the monitored object to normal operation. This real-time monitoring method fully exploits the statistical characteristics of real-time data, adaptively determines the operating status, and promptly adjusts the control parameters, thereby improving the degree of automation of monitoring.

[0005] For example, the invention patent announcement with the announcement number CN117806246A discloses a workstation control mode conversion method and system, which provides a workstation control mode conversion method and system, belonging to the field of electrical equipment control technology. Specifically, it includes collecting historical command data, intercepting and obtaining designated command data according to each control mode of the workstation; iteratively cross-validating the designated command data to obtain the credibility score of each designated command data; calculating the correlation score between each designated command data and the corresponding control mode based on the credibility score; combining the correlation score with the configuration information of multiple working modules controlled by the workstation, and performing deep learning to obtain an identification model of the designated command data and the control mode; according to the input control mode requirements, identifying the current operation status of the workstation based on the identification model, obtaining control mode flow data, and switching the control mode of the workstation through the control mode flow data. The present invention can ensure the accuracy of the corresponding control commands and the safety of electrical equipment when the workstation control mode is converted.

[0006] The above disclosed technical solutions have at least the following technical problems:

[0007] The use of fixed strategies for fault identification makes it impossible to dynamically select the optimal model combination based on real-time scenarios, resulting in a high misjudgment rate under complex working conditions. The dynamic changes in the virtual circuit topology are ignored, and key features such as loop redundancy and signal path complexity are not incorporated into the decision-making basis. The lack of a closed-loop feedback optimization mechanism makes it impossible to continuously update the model strategy based on actual repair results, making it difficult to cope with the challenges brought by new fault modes and the evolution of power grid structures.

[0008] In view of the above problems, the present invention proposes a solution. Summary of the Invention

[0009] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an intelligent substation fault identification method and system based on virtual loop topology, which solves the problem of balancing the misjudgment rate and timeliness caused by dynamic topology changes in the traditional single model through a closed-loop feedback optimization mechanism and a model control switching method.

[0010] To achieve the above object, the present invention provides the following technical solutions:

[0011] An intelligent substation fault identification method based on virtual loop topology includes the following steps: obtaining a first model and a second model, and constructing an identification mode, wherein the identification mode includes independent identification and combined identification; identifying substation faults through the identification mode, and regulating substation equipment nodes according to the identification results; the identification mode is switched through a control switching system; the control switching system is obtained by obtaining model identification features and constructing it based on random forest, and the control switching system includes verification of the identification results and system update.

[0012] In a preferred embodiment, the model recognition features include virtual loop signal data, the accuracy of the first model and the second model, and the loop topology complexity obtained based on graph theory analysis; the first model includes a rule base model constructed based on expert rules, and the second model includes a machine learning model constructed based on deep learning.

[0013] In a preferred embodiment, the first model is specifically obtained by: converting the physical connection of the substation into a virtual circuit topology structure; constructing a rule base according to the first rule, the first rule including topology constraint rules, combination mapping rules and conflict resolution rules; parsing the virtual circuit topology structure, traversing the fault library based on forward chain reasoning according to the topology state and virtual circuit signal data, and matching and outputting the fault cause.

[0014] In a preferred embodiment, the second model is specifically obtained by: obtaining historical fault recording data as training data, constructing an LSTM layer and a GNN layer; extracting the time series features of the historical fault recording data based on the LSTM layer, and performing topological analysis on the historical fault recording data based on the GNN layer to obtain topological features; fusing the time series features and topological features to obtain the second model, and the fully connected layer outputs the cause of the fault.

[0015] In a preferred embodiment, the recognition mode is switched through a control switching system, and the specific method is: obtaining historical fault data as training data, taking the highest accuracy and the shortest recognition time as decision-making goals, marking the historical fault recognition decision label as the first model independent recognition mode or the second model independent recognition mode, and constructing a control switching system based on random forest, wherein the historical fault data includes model recognition features, historical fault causes and historical verification conclusions obtained according to simulation; obtaining virtual loop signal data and loop topology complexity in real time based on the SCADA system and PMU as input of the control switching system; outputting the optimal recognition mode through the control switching system, canceling the recognition time constraint when manual intervention occurs, and limiting the recognition mode to a combined recognition mode.

[0016] In a preferred embodiment, the verification of the identification results and the update of the system are specifically as follows: virtual nodes are supplemented and replaced in the substation virtual circuit topology according to the identification results; based on the comparison and verification of the supplementation and replacement results with the preset normal virtual circuit topology, it is determined whether the identification is correct, and the verification conclusion is obtained in combination with the identification time; the verification also includes if the identification error occurs for a preset number of consecutive times, limiting the identification mode to the first model independent identification mode, and alerting manual intervention; adding the model identification features, identification results and verification conclusions to the historical fault data, and training and updating the control switching system.

[0017] In a preferred embodiment, the loop topology complexity is specifically obtained by: constructing a virtual loop topology relationship diagram of the device nodes based on the graph theory model and the substation SCD file to obtain the number of virtual nodes, the number of signal paths and the graph density; calculating the path redundancy index of the node according to the network redundancy, quantifying the heterogeneity index of the substation equipment type distribution according to the entropy method, determining the weight of each dimension according to the hierarchical analysis method, normalizing the path redundancy index and the heterogeneity index, constructing a linear weighted comprehensive scoring model, and outputting the loop topology complexity.

[0018] In a preferred embodiment, the recognition method of the combined recognition pattern includes: adding a preset number of neurons to the output layer of the second model, wherein the neurons correspond to the confidence of independent fault causes; selecting the second model for recognition and outputting several fault causes with the highest confidence; constructing a first fault set, wherein the first fault set includes several fault causes output by the second model; selecting the first model, traversing the first fault set, matching and outputting the fault causes, and restoring the output layer of the second model after the fault causes are output.

[0019] In a preferred embodiment, regulating the substation equipment nodes according to the identification results is specifically: after determining that the identification is correct, converting the virtual circuit topology structure into a substation physical connection, and repairing and replacing the equipment nodes of the substation physical connection.

[0020] A system for an intelligent substation fault identification method based on virtual loop topology includes an identification module, a control switching module, a verification and update module, and a regulation module: the identification module is used to obtain a first model and a second model, and construct an identification mode, and identify substation faults through the identification mode, and the identification mode includes independent identification and combined identification; the control switching module is used to switch the identification mode through a control switching system, and the control switching system is obtained by obtaining model identification features and constructing it based on random forest; the verification and update module is used to verify the identification results and update the system; the regulation module is used to regulate the substation equipment nodes according to the identification results.

[0021] The technical effects and advantages of the intelligent substation fault identification method and system based on virtual circuit topology of the present invention are as follows:

[0022] The present invention builds a fast model by building a fault library and fusing expert rules, builds a slow model based on deep learning, and forms recognition patterns independently or in combination, and dynamically selects the optimal recognition pattern in combination with random forest, thereby achieving a double improvement in the accuracy and timeliness of virtual circuit fault recognition; further, based on the quantitative evaluation of topological complexity, the adaptability in complex topological scenarios is enhanced; through the repair verification feedback loop and continuous iterative update of the model, it helps to build a self-optimizing fault diagnosis system, and effectively solves the problems of high misjudgment rate, rigid strategy and strong manual dependence caused by dynamic topological changes in traditional single models. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic flow chart of a method for identifying faults in an intelligent substation based on virtual circuit topology provided in an embodiment of the present invention.

[0024] Figure 2 A schematic diagram of the system structure of a smart substation fault identification method based on virtual circuit topology provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0026] Example 1, Figure 1 The present invention provides a method for identifying faults in an intelligent substation based on a virtual circuit topology, comprising the following steps:

[0027] S1, obtaining a first model and a second model, and constructing a recognition mode, wherein the recognition mode includes independent recognition and combined recognition;

[0028] S2, identifying substation faults through recognition patterns, and regulating substation equipment nodes according to the recognition results; the recognition patterns are switched through a control switching system; the control switching system is obtained by acquiring model recognition features and constructing it based on random forests, and the control switching system includes verification of recognition results and system updates.

[0029] This embodiment builds a fast model by fusing expert rules with a fault library, builds a slow model based on deep learning, and forms recognition patterns independently or in combination. It combines random forests to dynamically select the optimal recognition pattern, thereby achieving both improved accuracy and timeliness in virtual circuit fault recognition. Furthermore, based on quantitative evaluation of topological complexity, it enhances adaptability in complex topological scenarios. Through the repair verification feedback loop and continuous iterative updates of the model, it helps to build a self-optimizing fault diagnosis system, effectively solving the problems of high misjudgment rate, rigid strategy, and strong manual dependence caused by dynamic topological changes in traditional single models.

[0030] S1, obtaining a first model and a second model, and constructing a recognition mode, wherein the recognition mode includes independent recognition and combined recognition.

[0031] In this embodiment, the first model includes a rule base model constructed based on expert rules, and the second model includes a machine learning model constructed based on deep learning.

[0032] In this embodiment, the first model is specifically obtained by:

[0033] Convert substation physical connections into virtual circuit topology;

[0034] Building a rule base according to the first rule, the first rule includes a topology constraint rule, a combination mapping rule and a conflict resolution rule;

[0035] Analyze the virtual circuit topology structure, traverse the fault library based on forward chain reasoning according to the topology status and virtual circuit signal data, and match and output the fault cause.

[0036] In this embodiment, the topology constraint rules include virtual circuit connectivity; the combination mapping rules include the logical matching relationship between multiple source signals and fault types; and the conflict resolution rules include arbitration of conflicting conclusions caused by multiple rule triggering.

[0037] It should be noted that topology constraint rules are used to verify the physical legitimacy of the virtual circuit structure. For example, differential protection must form a closed loop with the corresponding CT / PT. If a loop break or illegal bypass is detected, the fault judgment will not be triggered. Combination mapping rules are used to establish the connection between signal combinations and faults. For example, when the signal combination of "differential protection action + circuit breaker tripping + regional voltage loss" occurs, it is mapped to "intra-region short circuit", while "overcurrent protection action + circuit breaker refusal to operate + voltage drop" points to "circuit breaker failure". Conflict resolution rules are used to arbitrate contradictory conclusions when multiple rules are triggered to ensure the uniqueness of the diagnostic results, such as priority sorting.

[0038] In this embodiment, the second model is specifically obtained by:

[0039] Obtain historical fault recording data as training data and build LSTM layer and GNN layer;

[0040] The LSTM layer is used to extract the temporal features of historical fault recording data, and the GNN layer is used to perform topological analysis on the historical fault recording data to obtain topological features.

[0041] The second model is obtained by fusing the timing features and topological features, and the fully connected layer outputs the fault cause.

[0042] It should be noted that the first model is suitable for rapid fault identification, but it is difficult to cope with fault analysis under complex topologies. The second model is suitable for more complex topology scenarios, but the processing time is longer.

[0043] S2, identifying substation faults through recognition patterns, and regulating substation equipment nodes according to the recognition results; the recognition patterns are switched through a control switching system; the control switching system is obtained by acquiring model recognition features and constructing it based on random forests, and the control switching system includes verification of recognition results and system updates.

[0044] In this embodiment, the model identification features include virtual loop signal data, the accuracy of the first model and the second model, and the loop topology complexity obtained according to graph theory analysis.

[0045] In this embodiment, the recognition mode is switched by controlling the switching system, specifically by:

[0046] Historical fault data is obtained as training data. With the highest accuracy and shortest recognition time as the decision-making goals, historical fault recognition decision labels are labeled as the first model independent recognition mode or the second model independent recognition mode. A control switching system is constructed based on random forest. The historical fault data includes model recognition features obtained through simulation, historical fault causes, and historical verification conclusions.

[0047] Based on the SCADA system and PMU, the virtual loop signal data and loop topology complexity are obtained in real time as input to the control switching system;

[0048] The optimal recognition mode is output by controlling the switching system. When human intervention occurs, the recognition time constraint is canceled and the recognition mode is limited to a combined recognition mode.

[0049] In this embodiment, the loop topology complexity is specifically obtained by:

[0050] Based on the graph theory model and the substation SCD file, a virtual circuit topology diagram of the equipment nodes is constructed to obtain the number of virtual nodes, the number of signal paths and the graph density;

[0051] The path redundancy index of the node is calculated according to the network redundancy. The heterogeneity index of the substation equipment type distribution is quantified according to the entropy method. The weight of each dimension is determined according to the hierarchical analysis method. The path redundancy index and heterogeneity index are normalized, and a linear weighted comprehensive scoring model is constructed to output the loop topology complexity.

[0052] It should be noted that the virtual circuit signal data includes protection action signals, equipment status signals, electrical parameter sampling and control command signals.

[0053] A Supervisory Control and Data Acquisition (SCADA) system is an automation system used for real-time monitoring and control of industrial equipment and infrastructure. It collects data such as temperature, pressure, and flow through remote terminal units (RTUs) or programmable logic controllers (PLCs) distributed across the field and transmits this data to a central control room via a communications network. SCADA systems allow operators to remotely monitor equipment status, adjust parameters, or trigger control commands. They are widely used in fields such as electricity, water, oil and gas, and transportation, and are particularly well-suited for distributed systems requiring centralized management. While they offer strong real-time performance, wide coverage, and a user-friendly human-computer interface, they focus on steady-state monitoring and have limited responsiveness to dynamic processes.

[0054] The synchronized phasor measurement unit (PMU) is the core equipment of the wide-area measurement system (WAMS), primarily used for high-precision dynamic monitoring of power systems. Using synchronized phasor technology, combined with GPS clock signals, the amplitude, phase, and frequency of electrical quantities such as bus voltage and line current are collected at high frequency, and the data is synchronously uploaded to the master station. The PMU can capture system transient processes such as short circuits and oscillations, providing real-time dynamic information for power dispatch, supporting grid stability analysis, fault location, and optimized control strategies. Compared to traditional SCADA systems, the PMU has higher time synchronization accuracy and data refresh rate, but is more expensive. It is primarily used for advanced monitoring and protection of high-voltage transmission networks and large power systems.

[0055] Random forest is an ensemble learning algorithm based on decision trees. It improves the generalization ability of the model by constructing multiple decision trees and integrating their prediction results. Its core advantage lies in effectively reducing the risk of overfitting while maintaining the model's high efficiency through self-sampling and random feature selection. Models built based on random forests have a remarkable lightweight feature: the natural hierarchical structure of the decision tree itself has low computational complexity, and the parallel training mechanism further improves efficiency; the model can achieve stable performance without complex parameter tuning, and the final integrated results can be quickly output through simple voting or averaging, without the need for additional deep calculations. This lightweightness enables it to perform well in resource-constrained environments, enabling rapid training and inference while achieving high-precision predictions with a smaller model size, making it an ideal choice for balancing performance and efficiency.

[0056] The Substation Configuration Description (SCD) file is a standard core configuration file for substation automation systems. It describes the communication parameters, data model, logical node configuration, and network communication relationships of all intelligent electronic devices (IEDs) within the substation. It contains key information such as the substation's physical structure, device function definitions, GOOSE / SV communication configuration, data set mapping, and report control blocks.

[0057] In this embodiment, the method for identifying the combined identification pattern includes:

[0058] Adding a preset number of neurons to the output layer of the second model, wherein the neurons correspond to confidences of independent fault causes;

[0059] Select the second model for identification and output several fault causes with the highest confidence level;

[0060] Constructing a first fault set, where the first fault set includes several fault causes output by the second model;

[0061] Select the first model, traverse the first fault set, match and output the fault cause, and restore the output layer of the second model after the fault cause is output.

[0062] In this embodiment, the verification of the recognition result and the update of the system are specifically as follows:

[0063] According to the identification results, virtual nodes are supplemented and replaced in the substation virtual circuit topology;

[0064] Based on the comparison and verification of the supplementation and replacement results with the preset normal virtual circuit topology, it is determined whether the recognition is correct, and a verification conclusion is obtained in combination with the recognition time; the verification also includes limiting the recognition mode to the first model independent recognition mode if the recognition error occurs for a preset number of consecutive times, and alerting manual intervention;

[0065] The model recognition features, recognition results and verification conclusions are added to the historical fault data to train and update the control switching system.

[0066] In this embodiment, the verification conclusion includes recognition correctness and recognition time.

[0067] In this embodiment, the substation equipment nodes are regulated according to the identification results, specifically:

[0068] After confirming that the identification is correct, the virtual circuit topology is converted into a physical connection of the substation, and the device nodes of the physical connection of the substation are repaired and replaced.

[0069] Example 2, Figure 2 The present invention provides a system for a fault identification method for an intelligent substation based on a virtual circuit topology, including an identification module, a control switching module, a verification and update module, and a control module.

[0070] an identification module, configured to obtain the first model and the second model, and construct an identification pattern, and identify the substation fault through the identification pattern, wherein the identification pattern includes independent identification and combined identification;

[0071] A control switching module is used to switch the recognition mode by controlling the switching system, wherein the control switching system is obtained by obtaining the recognition features of the model and constructing it based on the random forest;

[0072] Verification and update module, used for verification of identification results and system updates;

[0073] The control module is used to control the substation equipment nodes according to the identification results.

[0074] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0075] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0076] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. 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 beyond the scope of this application.

[0077] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0078] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0079] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for identifying faults in intelligent substations based on virtual circuit topology, characterized in that: The following steps are involved: Obtaining a first model and a second model, and constructing a recognition mode, wherein the recognition mode includes independent recognition and combined recognition, the first model includes a rule base model constructed based on expert rules, and the second model includes a machine learning model constructed based on deep learning; Identify substation faults through recognition patterns and regulate substation equipment nodes based on the recognition results; the recognition patterns are switched through a control switching system; the control switching system is constructed by acquiring model recognition features and based on random forests. The control switching system includes taking virtual circuit signal data and circuit topology complexity as input, outputting the optimal recognition pattern, and also includes verification of the recognition results and system updates. The identification mode is switched by controlling the switching system, specifically in the following manner: Historical fault data is obtained as training data. With the highest accuracy and shortest recognition time as the decision-making goals, historical fault recognition decision labels are labeled as the first model independent recognition mode or the second model independent recognition mode. A control switching system is constructed based on random forest. The historical fault data includes model recognition features obtained through simulation, historical fault causes, and historical verification conclusions. Based on the SCADA system and PMU, the virtual loop signal data and loop topology complexity are obtained in real time as input to the control switching system; The optimal recognition mode is output by controlling the switching system. When human intervention occurs, the recognition time constraint is canceled and the recognition mode is limited to a combined recognition mode.

2. The method for identifying faults in a smart substation based on virtual circuit topology according to claim 1, characterized in that: The model identification features include virtual loop signal data, the accuracy of the first model and the second model, and the loop topology complexity obtained according to graph theory analysis.

3. The method for identifying faults in a smart substation based on virtual circuit topology according to claim 2, characterized in that: The specific method for obtaining the first model is: Convert substation physical connections into virtual circuit topology; Building a rule base according to the first rule, the first rule includes a topology constraint rule, a combination mapping rule and a conflict resolution rule; Analyze the virtual circuit topology structure, traverse the fault library based on forward chain reasoning according to the topology status and virtual circuit signal data, and match and output the fault cause.

4. The method for identifying faults in a smart substation based on virtual circuit topology according to claim 3, characterized in that: The specific method for obtaining the second model is: Obtain historical fault recording data as training data and build LSTM layer and GNN layer; The LSTM layer is used to extract the temporal features of historical fault recording data, and the GNN layer is used to perform topological analysis on the historical fault recording data to obtain topological features. The second model is obtained by fusing the timing features and topological features, and the fully connected layer outputs the fault cause.

5. The method for identifying faults in a smart substation based on virtual circuit topology according to claim 4, characterized in that: The verification of the recognition results and the update of the system are specifically as follows: According to the identification results, virtual nodes are supplemented and replaced in the substation virtual circuit topology; Based on the comparison and verification of the supplementation and replacement results with the preset normal virtual circuit topology, it is determined whether the recognition is correct, and a verification conclusion is obtained in combination with the recognition time; the verification also includes limiting the recognition mode to the first model independent recognition mode if the recognition error occurs for a preset number of consecutive times, and alerting manual intervention; The model recognition features, recognition results and verification conclusions are added to the historical fault data to train and update the control switching system.

6. The method for identifying faults in a smart substation based on virtual circuit topology according to claim 5, characterized in that: The loop topology complexity is specifically obtained as follows: Based on the graph theory model and the substation SCD file, a virtual circuit topology diagram of the equipment nodes is constructed to obtain the number of virtual nodes, the number of signal paths and the graph density; The path redundancy index of the node is calculated according to the network redundancy. The heterogeneity index of the substation equipment type distribution is quantified according to the entropy method. The weight of each dimension is determined according to the hierarchical analysis method. The path redundancy index and heterogeneity index are normalized, and a linear weighted comprehensive scoring model is constructed to output the loop topology complexity.

7. The method for identifying faults in a smart substation based on virtual circuit topology according to claim 6, characterized in that: The recognition method of the combined recognition mode includes: Adding a preset number of neurons to the output layer of the second model, wherein the neurons correspond to confidences of independent fault causes; Select the second model for identification and output several fault causes with the highest confidence level; Constructing a first fault set, where the first fault set includes several fault causes output by the second model; Select the first model, traverse the first fault set, match and output the fault cause, and restore the output layer of the second model after the fault cause is output.

8. The method for identifying faults in a smart substation based on virtual circuit topology according to claim 7, characterized in that: The control of the substation equipment nodes according to the identification results is specifically as follows: After confirming that the identification is correct, the virtual circuit topology is converted into a physical connection of the substation, and the device nodes of the physical connection of the substation are repaired and replaced.

9. A system using the intelligent substation fault identification method based on virtual circuit topology according to any one of claims 1 to 8, characterized in that: Including identification module, control switching module, verification and update module and regulation module: an identification module, configured to obtain the first model and the second model, and construct an identification pattern, and identify the substation fault through the identification pattern, wherein the identification pattern includes independent identification and combined identification; A control switching module is used to switch the recognition mode by controlling the switching system, wherein the control switching system is obtained by obtaining the recognition features of the model and constructing it based on the random forest; Verification and update module, used for verification of identification results and system updates; The control module is used to control the substation equipment nodes according to the identification results.

Images