Power grid operation weak point identification method and system based on transmission and distribution network topological structure

Abstract

The invention discloses a power grid operation weak point identification method and system based on a transmission and distribution network topological structure, and relates to the technical field of power grid operation, and the method effectively reduces the calculation cost, reduces the model complexity through a graph theory and a network simplification technology, optimizes the use of calculation resources, improves the data processing efficiency, and reduces the calculation cost. Therefore, accurate identification under limited calculation performance is realized; data quality indexes are read and evaluated, and data enhancement and dimensionless processing are performed, so that the integrity and accuracy of data are improved, and the stability of the system during data quality fluctuation is ensured; a deep learning algorithm is adopted, a model training process is optimized through cross validation and hyper-parameter tuning, and supervised learning is performed by using fault and operation historical data, so that dependence on a large amount of labeled data and computing resources is reduced, and efficient and accurate power grid operation weak point identification is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid operation, and in particular to a method and system for identifying weak points in power grid operation based on a transmission and distribution network topology. Background Art

[0002] The grid operation weak point identification method aims to identify the weak links in the grid that may cause system instability or failure. As the scale and complexity of the grid increase, the traditional grid analysis method gradually exposes its limitations. To solve this problem, researchers and engineers have introduced advanced analysis tools and technologies, including big data analysis, machine learning and deep learning algorithms. These technologies can extract valuable information from massive grid operation data, thereby more accurately identifying potential risk points and system bottlenecks.

[0003] Although the grid operation weak point identification technology has many advantages, it also has some technical disadvantages when applied to the subsequent planning and optimization of the grid;

[0004] 1. High computing cost. Identifying weak points in power grid operation requires processing a large amount of real-time and historical data, which places high demands on computing performance.

[0005] 2. Insufficient data quality requirements. Existing methods have high requirements for data integrity and accuracy. If the collected data is missing, wrong or noisy, it will directly affect the accuracy and reliability of the recognition results. Fluctuations in data quality will have a negative impact on the stability of the system.

[0006] 3. Model training is complex. Although advanced technologies such as deep learning have significantly improved recognition accuracy, the model training and tuning process is complex and requires a large amount of labeled data and computing resources. Summary of the invention

[0007] In view of the above-mentioned problems, the present invention is proposed.

[0008] Therefore, the technical problem solved by the present invention is that the existing recognition methods have optimization problems such as high cost, poor data quality, and huge amount of calculation.

[0009] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network, comprising:

[0010] Collect and pre-process the transmission and distribution network topology data of the power grid, establish the power grid topology model, and use graph theory and network simplification technology to simplify the model;

[0011] Read data quality information from monitoring devices and sensors, calculate and evaluate the data quality information, and adjust the data quality information based on the evaluation results;

[0012] Based on the simplified grid topology model and the adjusted data quality information, key features are extracted and the feature representation method is used to convert the topology information into feature vectors that can be processed by the model.

[0013] Use deep learning algorithms to train models on the extracted feature matrix; adjust model parameters through cross-validation and hyperparameter tuning;

[0014] Deploy the trained model in the real-time monitoring system of the power grid, and analyze and process the real-time collected operation data; calculate and obtain the operation defects through the prediction results of the model;

[0015] Based on the assessment results, the operational weaknesses of the power grid can be identified in real time, early warnings and recommended measures can be provided, and fault handling and prevention can be carried out.

[0016] As a preferred solution of the method for identifying weak points in power grid operation based on the topological structure of the power transmission and distribution network described in the present invention, wherein: the collecting and preprocessing of the topological structure data of the power transmission and distribution network includes collecting the topological structure-related data of the power transmission and distribution network from the GIS system, the SCADA system, the equipment manual and the historical records, and constructing the data set of the power transmission and distribution network after data cleaning and dimensionless processing;

[0017] Based on the collected data, the node and line models of the power grid are constructed, and the electrical parameters of the equipment are entered;

[0018] The power grid topology model includes the connection relationship and electrical parameters of each node, line, and device;

[0019] The use of graph theory and network simplification technology to simplify the model includes identifying and retaining key nodes and lines, and deleting redundant parts, thereby forming a simplified topological model for reflecting the actual situation of the power grid.

[0020] As a preferred solution of the method for identifying weak points in power grid operation based on the topology of the transmission and distribution network described in the present invention, wherein: the data quality information includes data missing rate, data calibration frequency, data offset frequency and data redundancy rate;

[0021] The calculation and evaluation of the data quality information includes performing dimensionless processing on the data quality information and then calculating and obtaining the quality index Zlzs:

[0022]

[0023] Among them, Lqs represents the data missing rate, Ljz represents the data calibration frequency, Lpy represents the data offset frequency, and Lry represents the data redundancy rate;

[0024] By comparing the preset quality threshold Q with the quality index Zlzs, the evaluation result is generated:

[0025] If the preset quality threshold Q≤quality index Zlzs, it means that the data quality is qualified, and the first quality result is generated at this time; if the preset quality threshold Q>quality index Zlzs, it means that the data quality is unqualified, and the second quality result is generated at this time;

[0026] When the quality index Zlzs is the first quality result, it directly enters the subsequent analysis and processing stage without additional adjustment; when the quality index Zlzs is the second quality result, the data quality is adjusted at this time, including adjusting the data acquisition frequency, adjusting the data calibration frequency, adjusting the data synchronization mechanism, performing data merging processing, and performing data interpolation and data synthesis;

[0027] After the data quality adjustment is completed, the evaluation quality index Zlzs is calculated again until the quality index Zlzs is higher than the preset quality threshold Q.

[0028] As a preferred solution of the method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network described in the present invention, the key features include the electrical characteristics of the nodes, the load conditions of the lines and the fault history records;

[0029] The electrical characteristics of the node include voltage, current and power factor; the load condition of the line includes load current, line resistance and reactance; the fault history record includes the number of faults, fault type and fault time of each node and line;

[0030] By extracting the features of the voltage, current and power factor of each node, a node feature set is formed;

[0031] By extracting the characteristics of the load current, resistance and reactance of each line, a line feature set is formed;

[0032] By counting the number of faults, fault types and fault time of each node and line, feature extraction is performed to form a fault feature set;

[0033] The characteristic vector includes: the characteristic representation of the node characteristic vector is: the electrical characteristic parameter of each node is represented as a characteristic vector V a =[v a ,i a , pf a ], where v a represents the voltage at node a, I a represents the current at node a, pf a represents the power factor of node a;

[0034] The characteristic representation of the line characteristic vector is: The load condition parameters of each line are represented as the characteristic vector I b =[i b , r b , x b ], where i b represents the load current of line b, r b represents the resistance of line b, x b represents the reactance of line b;

[0035] The characteristic representation of the fault feature vector is: The fault history record of each node and line is represented as the feature vector F c =[f c , t c , n c ], where f c represents the number of failures of line c, t c Indicates the fault type of line c, n c represents the fault time of line c;

[0036] The node feature vector, line feature vector and fault feature vector are combined with the topological structure information to generate a comprehensive feature vector; the feature vectors of nodes, lines and faults are connected according to the grid topology using graph theory to construct an overall feature graph G=(D, E), where D is the node set and E is the line set;

[0037] Format the comprehensive feature vector; and normalize the feature vector to generate the final model-processable feature vector set X=[V, I, F];

[0038] V = [V1, V2, …, V q ]; I = [I1, I2, …, I m ]; F = [F1, F2, …, F e ];

[0039] Among them, V q represents the feature vector of the qth node, I m represents the feature vector of the mth line, F e represents the fault feature vector of the e-th node or line, V represents the feature vector set of the node, I represents the feature vector set of the line, and F represents the fault feature vector set.

[0040] As a preferred solution of the method for identifying weak points in power grid operation based on the topology of the transmission and distribution network described in the present invention, wherein: the deep learning algorithm includes setting initial model parameters by selecting the convolutional neural network structure and the number of layers;

[0041] Use the extracted feature matrix as training data, use the back propagation algorithm to train the model, and update the model parameters to minimize the loss function;

[0042] During the training process, cross-validation techniques were applied to divide the data into training sets and validation sets to evaluate the generalization ability of the model;

[0043] Adjust learning rate, batch size, and regularization parameters through hyperparameter tuning;

[0044] Use existing fault and operation history data for supervised learning;

[0045] Through multiple rounds of training and verification, the model parameters are continuously adjusted until the model reaches the expected accuracy and stability;

[0046] The trained model is applied to actual power grid operation data to identify and predict weak points in the power grid.

[0047] As a preferred solution of the method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network described in the present invention, the node feature vector V is extracted. d =[v d ,i d , pf d ], line feature vector I j =[ i j , r j , x j ] and fault feature vector F k =[f k , t k , n k ], and after dimensionless processing, the operating defect value Yxqx is calculated:

[0048]

[0049] Where e represents the number of node feature vectors, r represents the number of line feature vectors, and t represents the number of fault feature vectors; V d Represents the value of the dth node feature vector, I j represents the value of the jth line feature vector, F k represents the value of the kth fault feature vector;

[0050] By comparing the preset defect threshold W with the operating defect value Yxqx, the result is generated:

[0051] If the operating defect value Yxqx is less than the defect threshold W, it indicates that the defect is normal or there is no defect, and a first defect result is generated; if the operating defect value Yxqx is greater than or equal to the defect threshold W, it indicates that the defect is abnormal, and a second defect result is generated to identify the operating weaknesses of the power grid.

[0052] As a preferred solution of the method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network described in the present invention, wherein: the operation weak points of the power grid are identified in real time according to the evaluation results, early warning and suggested measures are provided, and fault processing and prevention are performed, including: when the second defect result is generated, relevant node and line information is recorded, and specific areas or equipment with problems in the power grid are determined;

[0053] Establish an early warning mechanism. When a weak point in operation is detected, the system alarm will notify the relevant operation and maintenance personnel and management personnel, and provide detailed early warning information, including the specific location, type, possible cause and recommended treatment measures of the weak point; the recommended treatment measures include load adjustment, equipment maintenance, line adjustment and modification, and configuration adjustment;

[0054] Locate the fault point, combine historical data and model prediction results, determine the cause of the fault, and implement emergency repair plans; at the same time, record the fault handling process and results, and update the fault database; formulate a detailed preventive maintenance plan, including inspection and maintenance of key equipment and lines within a fixed period.

[0055] A power grid operation weak point identification system based on the transmission and distribution network topology structure using any method as described in the present invention, wherein:

[0056] The acquisition unit collects and pre-processes the topological structure data of the power transmission and distribution network, establishes a power grid topology model, and uses graph theory and network simplification technology to simplify the model;

[0057] An adjustment unit, which reads data quality information from monitoring equipment and sensors, calculates and evaluates the data quality information, and adjusts the data quality information according to the evaluation results;

[0058] The processing unit extracts key features based on the simplified power grid topology model and the adjusted data quality information, and uses a feature representation method to convert the topology structure information into a feature vector that can be processed by the model;

[0059] The model optimization unit uses deep learning algorithms to train the model on the extracted feature matrix; it adjusts the model parameters through cross-validation and hyperparameter tuning;

[0060] The computing unit deploys the trained model in the real-time monitoring system of the power grid and analyzes and processes the real-time collected operation data; it calculates and obtains the operation defects through the prediction results of the model;

[0061] The output unit identifies the operational weaknesses of the power grid in real time based on the evaluation results, provides early warnings and recommended measures, and performs fault handling and prevention.

[0062] A computer device comprises: a memory and a processor; the memory stores a computer program, wherein: the processor implements the steps of any one of the methods of the present invention when executing the computer program.

[0063] A computer-readable storage medium stores a computer program, wherein: when the computer program is executed by a processor, the steps of any one of the methods of the present invention are implemented.

[0064] Beneficial effects of the present invention: The method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network provided by the present invention effectively reduces the complexity of the model by applying graph theory and network simplification technology; by identifying and retaining key nodes and lines and deleting redundant parts, the established power grid topology model is more concise and the amount of calculation is reduced; in addition, by combining the deep learning algorithm and the efficient representation method of the feature vector, the utilization of computing resources is optimized, the efficiency of data processing and analysis is improved, and thus the computing cost of the system is reduced. Finally, the goal of accurately and quickly identifying weak points in power grid operation under limited computing performance conditions is achieved. By reading data quality related information from monitoring equipment and sensors, and performing comprehensive computational evaluation and adjustment on data quality, the integrity and accuracy of the data are ensured; specifically, the monitoring and adjustment of key indicators such as data missing rate, data calibration frequency, data offset frequency and data redundancy rate are included; by performing dimensionless processing and data enhancement technology on the data, the data quality is improved, the influence of data noise and errors is reduced, thereby improving the accuracy and reliability of the identification results, and ensuring that the system can still maintain stable operation when the data quality fluctuates. A deep learning algorithm was adopted, and the model training process was optimized through cross-validation and hyperparameter tuning techniques. Existing fault and operation history data were used for supervised learning to improve the recognition accuracy of the model. At the same time, the normalization and unified processing of feature vectors made the construction of the feature matrix more efficient and reduced the dependence on large amounts of labeled data and computing resources. Through multiple rounds of training and verification, the model parameters were continuously adjusted to ensure that the model achieved the expected accuracy and stability, thus overcoming the difficulty of complex model training and achieving efficient and accurate identification of weak points in power grid operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0066] Figure 1 An overall flow chart of a method for identifying weak points in power grid operation based on a transmission and distribution network topology structure provided in the first embodiment of the present invention. DETAILED DESCRIPTION

[0067] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0068] Example 1, reference Figure 1 , which is an embodiment of the present invention, provides a method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network, including: first, by applying graph theory and network simplification technology, the complexity of the power grid topology model is reduced and the calculation cost is reduced; second, the data quality is comprehensively evaluated and adjusted to improve the integrity and accuracy of the data and ensure the stability of the system when the data quality fluctuates; through feature extraction and feature representation methods, complex topological structure information is converted into feature vectors that can be processed by the model, thereby improving data processing efficiency; using a deep learning algorithm, through cross-validation and hyperparameter tuning, the model training process is optimized, and supervised learning is performed using fault and operation history data to improve identification accuracy; finally, the trained model is deployed in a real-time monitoring system to realize dynamic analysis and processing of real-time data, timely identify weak points in power grid operation, provide early warnings and recommended measures, effectively prevent and handle faults, and comprehensively improve the safety and reliability of the power grid.

[0069] Step 1: Collect and pre-process the transmission and distribution network topology data of the power grid, and establish a detailed power grid topology model, including the connection relationship and electrical parameters of each node, line and equipment; at the same time, use graph theory and network simplification technology to simplify the model.

[0070] Firstly, the data related to the topological structure of the transmission and distribution network is collected from the GIS system, SCADA system, equipment manuals and historical records, and after data cleaning and dimensionless processing, a transmission and distribution network data set is constructed. Secondly, based on the collected data, the node and line model of the power grid is constructed and entered into the electrical parameters of the equipment. Then, graph theory and network simplification technology are applied to identify and retain key nodes and lines, and delete redundant parts, thereby forming a simplified topological model that reflects the actual situation of the power grid.

[0071] In this embodiment, the node model includes various devices in the power grid, including transformers and switches, and records their electrical parameters, including voltage, current and power factor; the line model includes various lines in the power grid, and also records their electrical parameters, including resistance, reactance and load; these model data are input into the system to lay the foundation for topological analysis and feature extraction of the power grid.

[0072] By applying network simplification technology, key nodes and lines are identified and retained, and redundant or low-impact parts are deleted. Specific technologies include: key node identification, using network centrality indicators, including betweenness centrality and closeness centrality, to identify nodes that are critical to the operation of the power grid; line screening, based on the load and electrical parameters of the line, screening out the lines that have a greater impact on the stability of the power grid; redundancy removal, using network compression technology to remove nodes and lines that have little impact on the function of the power grid and simplify the network structure.

[0073] The simplified node and line information is integrated to form a topological model that is clearer and truly reflects the actual situation of the power grid. The structure of the model is further optimized, and through iterative adjustment and verification, it is ensured that it can accurately describe the operating status of the power grid and adapt to future data changes and expansion needs. This process helps to improve the accuracy of power grid analysis and weak point identification.

[0074] Step 2: Read data quality related information from monitoring equipment and sensors, calculate and evaluate the data quality related information, and adjust the data quality related information based on the evaluation results, including preprocessing and data enhancement of the collected data.

[0075] By reading the data quality related information, including data missing rate, data calibration frequency, data offset frequency and data redundancy rate, and performing dimensionless processing, the quality index Zlzs is calculated. The specific calculation formula is as follows:

[0076]

[0077] Wherein, Lqs represents the data missing rate, Ljz represents the data calibration frequency, Lpy represents the data offset frequency, and Lry represents the data redundancy rate.

[0078] By comparing the preset quality threshold Q with the quality index Zlzs, the following evaluation results are generated:

[0079] If the preset quality threshold Q ≤ the quality index Zlzs, it means that the data quality is qualified, and a first quality result is generated. If the preset quality threshold Q > the quality index Zlzs, it means that the data quality is unqualified, and a second quality result is generated.

[0080] According to the evaluation results of the quality index Zlzs, the following adjustments are made:

[0081] When the quality index Zlzs is the first quality result, it directly enters the subsequent analysis and processing stage without additional adjustment. When the quality index Zlzs is the second quality result, the data quality is adjusted at this time, including adjusting the data acquisition frequency, adjusting the data calibration frequency, adjusting the data synchronization mechanism, performing data merging processing, and performing data interpolation and data synthesis.

[0082] After the data quality adjustment is completed, the evaluation quality index Zlzs is calculated again until the quality index Zlzs is higher than the preset quality threshold Q.

[0083] In this embodiment, the data missing rate Lqs indicates how much proportion of data is lost or not recorded during the collection process. A high missing rate will affect the integrity of the data and lead to inaccurate analysis results. By monitoring and reducing the data missing rate Lqs, the quality of the data set can be improved, thereby ensuring the effectiveness of model training and power grid analysis.

[0084] Data calibration frequency Ljz refers to the frequency of data calibration and correction. Regular calibration helps correct the drift and error of sensors or measuring equipment and ensure the accuracy and consistency of data; the higher the data calibration frequency Ljz, the more accurate the data is, which in turn improves the predictive performance and reliability of the model.

[0085] The data offset frequency Lpy indicates the frequency of data deviation, that is, the degree of deviation of the data from the actual value; frequent offsets may lead to data inconsistency, affecting data quality and the accuracy of analysis results; controlling and reducing the data offset frequency Lpy helps improve data stability and the prediction accuracy of the model.

[0086] The data redundancy rate Lry refers to the proportion of duplicate records in a data set; a high redundancy rate may lead to inefficient data processing and analysis; moderate redundancy can sometimes provide backup and redundant verification, but excessive redundancy will waste storage resources and may introduce noise; reducing the data redundancy rate Lry helps improve the efficiency and accuracy of data processing.

[0087] When the quality index Zlzs is the second quality result, the data quality is adjusted at this time, and the specific implementation process is as follows:

[0088] Adjust data collection frequency: evaluate whether the current data collection time interval is too large or too small; if the collection frequency is too low, increase the collection frequency to obtain more intensive data; if the collection frequency is too high and generates too much redundant data, appropriately reduce the collection frequency to optimize data storage and processing efficiency; re-calibrate the quality index Zlzs after adjustment.

[0089] Adjust data calibration frequency: Check whether the data calibration mechanism is timely, especially for possible drift or error problems of sensors or measuring equipment; optimize the calibration time interval based on historical error trends to ensure a balance between data calibration timeliness and accuracy; regenerate data after calibration and verify the improvement of the quality index Zlzs.

[0090] Adjust the data synchronization mechanism: verify the timestamps of multi-source data to ensure the time synchronization between different data sources; if time offset or data loss is found, use the time alignment algorithm to correct it; recalculate the quality index Zlzs of the updated synchronized data.

[0091] Data merging and processing: De-duplicate and integrate duplicate or related data from multiple sources to ensure that only a unique version of the same object is stored; use weighted average, optimization algorithm and other technologies to fuse multi-source data to improve data accuracy and consistency; the merged data is used to re-evaluate Zlzs.

[0092] Data interpolation: For missing values ​​in the data collection process, linear interpolation, spline interpolation or a time series-based prediction model is used to fill in the missing parts; after the interpolation is completed, local verification is performed to ensure that the interpolation results do not introduce additional errors.

[0093] Data synthesis: In specific scenarios, multi-dimensional data is synthesized according to needs. For example, voltage and current data are calculated to generate power data to improve the depth of analysis. After data synthesis, Zlzs is recalculated to confirm the effect of the synthesis result on the overall quality improvement.

[0094] Quality adjustment measures make data more reliable, thereby improving the effectiveness of model training and prediction accuracy, reducing the risks caused by data defects, and avoiding data redundancy and inconsistency through appropriate adjustment measures such as data synchronization and synthesis, thereby improving the overall stability and operating efficiency of the system; when data quality meets the requirements, unnecessary additional adjustments and processing are avoided, saving computing resources and processing time.

[0095] Step 3: Based on the topological structure and preprocessed operating data, extract key features, including the electrical characteristics of the nodes, the load conditions of the lines, and the fault history records; and use the feature representation method to convert the topological structure information into feature vectors that can be processed by the model.

[0096] The pre-processed operation data is adjusted data quality information. The electrical characteristics of the node include voltage, current and power factor; the load condition of the line includes load current, line resistance and reactance; the fault history record includes the number of faults, fault type and fault time of each node and line.

[0097] By collecting electrical parameters such as voltage, current, power factor, etc. of each node, feature extraction is performed to form a node feature set; by collecting load parameters such as load current, resistance, reactance, etc. of each line, feature extraction is performed to form a line feature set; by counting the number of faults, fault type and fault time of each node and line, feature extraction is performed to form a fault feature set.

[0098] The characteristic representation of the node feature vector is: the electrical characteristic parameters of each node are represented as a feature vector V a =[v a ,i a , pf a , …], where v a represents the voltage at node a, I a represents the current at node a, pf a Represents the power factor of node a, and so on.

[0099] The characteristic representation of the line characteristic vector is: The load condition parameters of each line are represented as the characteristic vector I b =[i b , r b , x b , …], where i b represents the load current of line b, r b represents the resistance of line b, x b represents the reactance of line b, and so on.

[0100] The characteristic representation of the fault feature vector is: The fault history record of each node and line is represented as the feature vector F c =[f c , t c , n c , …], where f c represents the number of failures of line c, t c Indicates the fault type of line c, n c Indicates the fault time of line c, and so on.

[0101] The node feature vector, line feature vector and fault feature vector are combined with the topological structure information to generate a comprehensive feature vector. Using graph theory methods, the feature vectors of nodes, lines and faults are connected according to the power grid topology to construct an overall feature graph G=(D, E), where D is the node set and E is the line set.

[0102] The comprehensive feature vector is formatted and normalized to generate the final model-processable feature vector set X=[V, I, F].

[0103] V = [V1, V2, …, Vq ]; I = [I1, I2, …, I m ]; F = [F1, F2, …, F e ].

[0104] Among them, V q represents the feature vector of the qth node, I m represents the feature vector of the mth line, F e represents the fault feature vector of the e-th node or line, V represents the feature vector set of the node, I represents the feature vector set of the line, and F represents the fault feature vector set.

[0105] In this embodiment, the electrical characteristics of the nodes, the load conditions of the lines, and the fault history records are extracted in detail and converted into feature vectors. This comprehensive data collection ensures an in-depth understanding of the operation status of the power grid and provides a rich information basis for subsequent analysis; the node feature vector, line feature vector, and fault feature vector are combined with the power grid topology information to generate a comprehensive feature vector. This feature fusion effectively integrates data from different sources, so that the power grid model can fully reflect the operating status and potential problems of various aspects; the overall feature graph G=(V, E) is constructed through graph theory methods, and effective modeling of the power grid topology is achieved. This modeling method facilitates feature correlation analysis in complex power grid structures and provides a clearer view of power grid operation; through systematic feature extraction, fusion, and normalization, the stability and reliability of the model are significantly improved, and the interference of data noise on the results is reduced, thereby effectively improving the ability to identify weak points in the power grid and the accuracy of prediction.

[0106] Step 4: Use deep learning algorithms to train the model of the extracted feature matrix; adjust the model parameters through cross-validation and hyperparameter tuning; and use existing fault and operation history data for supervised learning.

[0107] The deep learning model is initialized, and the initial model parameters are set by selecting the convolutional neural network structure and number of layers. Then, the extracted feature matrix is ​​used as training data, and the model is trained using the back-propagation algorithm, and the model parameters are updated to minimize the loss function. During the training process, the cross-validation technique is applied to divide the data into training set and validation set to evaluate the generalization ability of the model. The learning rate, batch size and regularization parameters are adjusted through hyperparameter tuning. During the training process, supervised learning is performed using existing fault and operation history data. Through multiple rounds of training and verification, the model parameters are continuously adjusted until the model reaches the expected accuracy and stability. Finally, the trained model is applied to the actual power grid operation data to identify and predict the weak points of the power grid.

[0108] In this embodiment, by initializing the deep learning model and selecting a suitable convolutional neural network structure, the feature matrix is ​​used for training to effectively improve the accuracy of identifying weak points in the power grid. The back propagation algorithm and cross-validation technology ensure the stability and generalization ability of the model on the training and validation sets, and the hyperparameter tuning further optimizes the performance of the model. Finally, through multiple rounds of training and verification, the high accuracy and reliability of the model are ensured. After being applied to actual power grid operation data, this technical solution can accurately identify weak points in the power grid and improve the stability and safety of power grid operation.

[0109] Step 5: Deploy the trained model in the real-time monitoring system of the power grid, and analyze and process the real-time collected operation data; calculate the operation defect value Yxqx through the prediction results of the model and evaluate it.

[0110] By extracting the node feature vector V d =[v d ,i d , pf d ], line feature vector I j =[ i j , r j , x j ] and fault feature vector F k =[f k , t k , n k ], and after dimensionless processing, the operating defect value Yxqx is calculated:

[0111]

[0112] Where e represents the number of node feature vectors, r represents the number of line feature vectors, and t represents the number of fault feature vectors; V d Represents the value of the dth node feature vector, I j represents the value of the jth line feature vector, F k Represents the value of the kth fault feature vector.

[0113] By comparing the preset defect threshold W with the operating defect value Yxqx, the following results are generated:

[0114] If the operation defect value Yxqx < the defect threshold W, it means the defect is normal or there is no defect, and the first defect result is generated. If the operation defect value Yxqx ≥ the defect threshold W, it means the defect is abnormal, and the second defect result is generated, and the operation weak points of the power grid are identified.

[0115] In this embodiment, the calculation of the running defect value Yxqx is performed by extracting the node feature vector V d , line characteristic vector I jand the fault feature vector F k , and dimensionless processing is performed, which is helpful for comprehensive evaluation of the power grid operation status; the specific calculation formula provides a comprehensive defect value measurement indicator; by comparing with the preset defect threshold W, it can clearly distinguish between normal and abnormal operation states.

[0116] Step 6: Based on the assessment results, identify the operational weaknesses of the power grid in real time, provide early warnings and recommended measures, and perform fault handling and prevention.

[0117] According to the second defect result, after identifying the weak points of the power grid operation, early warning and recommended measures are issued, and fault prevention and treatment are carried out. The specific contents are as follows:

[0118] When the second defect result is generated, the relevant node and line information is recorded, and the specific area or equipment with problems in the power grid is determined; secondly, an early warning mechanism is established. When an operation weakness is detected, the relevant operation and maintenance personnel and management personnel are notified through system alarms, and detailed early warning information is provided, including the specific location, type, possible cause and recommended treatment measures of the weakness; recommended treatment measures include load adjustment, equipment maintenance, line adjustment and modification, and configuration adjustment.

[0119] Then locate the fault point, combine historical data and model prediction results, determine the cause of the fault, and formulate and implement an emergency repair plan; at the same time, record the fault handling process and results, and update the fault database; finally, formulate a detailed preventive maintenance plan, including inspection and maintenance of key equipment and lines within a fixed period.

[0120] By establishing an early warning mechanism, once an abnormal defect is identified, the system alarm can be used to promptly notify the operation and maintenance personnel, provide detailed warning information and recommended measures, and effectively reduce the occurrence of accidents; formulate emergency repair plans and record the fault handling process to improve fault handling efficiency; at the same time, through preventive maintenance plans, regular inspections and maintenance of key equipment enhance the overall stability and security of the power grid.

[0121] On the other hand, this embodiment also provides a power grid operation weak point identification system based on the transmission and distribution network topology structure, which includes:

[0122] The acquisition unit collects the transmission and distribution network topology data of the power grid and performs preprocessing, establishes the power grid topology model, and uses graph theory and network simplification technology to simplify the model.

[0123] The adjustment unit reads data quality information from monitoring equipment and sensors, calculates and evaluates the data quality information, and adjusts the data quality information based on the evaluation results.

[0124] The processing unit extracts key features based on the simplified power grid topology model and the adjusted data quality information, and adopts a feature representation method to convert the topology structure information into a feature vector that can be processed by the model.

[0125] The model optimization unit uses a deep learning algorithm to perform model training on the extracted feature matrix and adjusts model parameters through cross-validation and hyperparameter tuning.

[0126] The computing unit deploys the trained model in the real-time monitoring system of the power grid and analyzes and processes the operating data collected in real time; the operating defects are calculated through the prediction results of the model.

[0127] The output unit identifies the operational weaknesses of the power grid in real time based on the evaluation results, provides early warnings and recommended measures, and performs fault handling and prevention.

[0128] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0129] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0130] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0131] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0132] Embodiment 2 is an embodiment of the present invention, which provides a method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0133] Experimental methods

[0134] 1. Subjects:

[0135] Select actual data of power grid operation in a certain region, including:

[0136] Grid topology data (number of nodes: 120, number of lines: 200).

[0137] Historical operation data (time span: 1 year, sampling frequency: 10 minutes).

[0138] Historical fault records (total: 50 times, involving line faults, node overloads, and voltage instability).

[0139] 2. Experimental and control group settings:

[0140] Experimental group: The weak point identification method based on the topological structure of the transmission and distribution network was adopted.

[0141] Control group: The traditional method based on static fault tree and manual rule analysis was used.

[0142] 3. Experimental steps:

[0143] 1) Data preprocessing:

[0144] Simplified modeling of power grid topology.

[0145] Process the collected historical operation data, calculate the quality index Zlzs, and adjust the data quality until it reaches the preset threshold Q.

[0146] 2) Feature extraction and model training:

[0147] Experimental group: Feature vectors were extracted based on node, line, and fault history records, and a feature graph G=(V,E)G=(V,E)G=(V,E) was constructed. The deep learning model was trained and optimized.

[0148] Control group: Based on fault tree modeling, a rule set for building nodes and lines is constructed.

[0149] 3) Real-time running data test:

[0150] Input real-time operation data (120 hours) for analysis and record the accuracy of defect identification, response time and warning effect.

[0151] Quantitative indicators:

[0152] Defect recognition accuracy:

[0153]

[0154] Response time: The time from real-time operation data input to the system output of defect identification results, unit: seconds.

[0155] Early warning coverage:

[0156] Operational weak point positioning error: the position deviation between the weak point located by the system and the actual weak point, unit: kilometer.

[0157] Experimental results and analysis:

[0158] Please refer to Table 1 for the experimental results.

[0159] analyze:

[0160] Improved accuracy: The new method automatically extracts multi-dimensional features through a deep learning model, with a recognition rate 14.2% higher than that of the traditional method (based on rule analysis); Shortened response time: The simplified topological model and efficient algorithm significantly shortened the data analysis time and improved real-time performance; Improved warning coverage: The supervision learning mechanism that integrates feature extraction and historical data makes the warning capability more comprehensive; Reduced positioning error: The fusion of topological map modeling and feature map makes the positioning of weak points more accurate.

[0161] Table 1 Data comparison table

[0162] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network, characterized in that: include: Collect and pre-process the transmission and distribution network topology data of the power grid, establish the power grid topology model, and use graph theory and network simplification technology to simplify the model; Read data quality information from monitoring devices and sensors, calculate and evaluate the data quality information, and adjust the data quality information based on the evaluation results; Based on the simplified grid topology model and the adjusted data quality information, key features are extracted and the feature representation method is used to convert the topology information into feature vectors that can be processed by the model. Use deep learning algorithms to train models on the extracted feature matrix; adjust model parameters through cross-validation and hyperparameter tuning; Deploy the trained model in the real-time monitoring system of the power grid, and analyze and process the real-time collected operation data; calculate and obtain the operation defects through the prediction results of the model; Based on the assessment results, the weak points of the power grid operation can be identified in real time, early warning and recommended measures can be provided, and fault handling and prevention can be performed; The collecting and preprocessing of the transmission and distribution network topology data of the power grid includes collecting the transmission and distribution network topology related data from the GIS system, the SCADA system, the equipment manual and the historical records, and constructing the transmission and distribution network data set after data cleaning and dimensionless processing; Based on the collected data, the node and line models of the power grid are constructed, and the electrical parameters of the equipment are entered; The power grid topology model includes the connection relationship and electrical parameters of each node, line, and device; The use of graph theory and network simplification technology to simplify the model includes identifying and retaining key nodes and lines, and deleting redundant parts, thereby forming a simplified topological model for reflecting the actual situation of the power grid; The data quality information includes data missing rate, data calibration frequency, data offset frequency and data redundancy rate; The calculation and evaluation of the data quality information includes performing dimensionless processing on the data quality information and then calculating and obtaining the quality index Zlzs: ， Among them, Lqs represents the data missing rate, Ljz represents the data calibration frequency, Lpy represents the data offset frequency, and Lry represents the data redundancy rate; By comparing the preset quality threshold Q with the quality index Zlzs, the evaluation result is generated: If the preset quality threshold Q≤quality index Zlzs, it means that the data quality is qualified, and the first quality result is generated at this time; if the preset quality threshold Q>quality index Zlzs, it means that the data quality is unqualified, and the second quality result is generated at this time; When the quality index Zlzs is the first quality result, it directly enters the subsequent analysis and processing stage without additional adjustment; when the quality index Zlzs is the second quality result, the data quality is adjusted at this time, including adjusting the data acquisition frequency, adjusting the data calibration frequency, adjusting the data synchronization mechanism, performing data merging processing, and performing data interpolation and data synthesis; After the data quality adjustment is completed, the evaluation quality index Zlzs is calculated again until the quality index Zlzs is higher than the preset quality threshold Q.

2. The method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network according to claim 1, characterized in that: The key characteristics include electrical characteristics of the node, load conditions of the line and fault history; The electrical characteristics of the node include voltage, current and power factor; The load condition of the line includes load current, line resistance and reactance; The fault history records include the number of faults, fault types and fault time of each node and line; By extracting the features of the voltage, current and power factor of each node, a node feature set is formed; By extracting the characteristics of the load current, resistance and reactance of each line, a line feature set is formed; By counting the number of faults, fault types and fault time of each node and line, feature extraction is performed to form a fault feature set; The characteristic vector includes: the characteristic representation of the node characteristic vector is: the electrical characteristic parameter of each node is represented as a characteristic vector V a =[v a ,i a , pf a ], where v a represents the voltage at node a, I a represents the current at node a, pf a represents the power factor of node a; The characteristic representation of the line characteristic vector is: The load condition parameters of each line are represented as the characteristic vector I b =[ i b , r b , x b ], where i b represents the load current of line b, r b represents the resistance of line b, x b represents the reactance of line b; The characteristic representation of the fault feature vector is: The fault history record of each node and line is represented as the feature vector F c =[f c , t c , n c ], where f c represents the number of failures of line c, t c Indicates the fault type of line c, n c represents the fault time of line c; The node feature vector, line feature vector and fault feature vector are combined with the topological structure information to generate a comprehensive feature vector; the feature vectors of nodes, lines and faults are connected according to the grid topology using graph theory to construct an overall feature graph G=(D, E), where D is the node set and E is the line set; Format the comprehensive feature vector; and normalize the feature vector to generate the final model-processable feature vector set X=[V, I, F]; V=[V1,V2,…,V q ];I=[I1,I2,…,I m ];F=[F1,F2,…,F e ]; Among them, V q represents the feature vector of the qth node, I m represents the feature vector of the mth line, F e represents the fault feature vector of the e-th node or line, V represents the feature vector set of the node, I represents the feature vector set of the line, and F represents the fault feature vector set.

3. The method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network according to claim 2, characterized in that: The deep learning algorithm includes setting initial model parameters by selecting a convolutional neural network structure and number of layers; Use the extracted feature matrix as training data, use the back propagation algorithm to train the model, and update the model parameters to minimize the loss function; During the training process, cross-validation techniques were applied to divide the data into training sets and validation sets to evaluate the generalization ability of the model; Adjust learning rate, batch size, and regularization parameters through hyperparameter tuning; Use existing fault and operation history data for supervised learning; Through multiple rounds of training and verification, the model parameters are continuously adjusted until the model reaches the expected accuracy and stability; The trained model is applied to actual power grid operation data to identify and predict weak points in the power grid.

4. The method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network according to claim 3, characterized in that: The operation defects include extracting the node feature vector V d =[v d ,i d , pf d ], line characteristic vector I j =[ i j , r j , x j ] and fault feature vector F k =[f k , t k , n k ], and after dimensionless processing, the operating defect value Yxqx is calculated: ， Where e represents the number of node feature vectors, r represents the number of line feature vectors, and t represents the number of fault feature vectors; V d Represents the value of the dth node feature vector, I j represents the value of the jth line feature vector, F k represents the value of the kth fault feature vector; By comparing the preset defect threshold W with the operating defect value Yxqx, the result is generated: If the operating defect value Yxqx is less than the defect threshold W, it indicates that the defect is normal or there is no defect, and a first defect result is generated; if the operating defect value Yxqx is greater than or equal to the defect threshold W, it indicates that the defect is abnormal, and a second defect result is generated to identify the operating weaknesses of the power grid.

5. The method for identifying weak points in power grid operation based on the topological structure of the transmission and distribution network according to claim 4, characterized in that: The method of identifying the operational weaknesses of the power grid in real time according to the evaluation results, providing early warnings and recommended measures, and performing fault handling and prevention includes: when the second defect result is generated, recording relevant node and line information, and determining the specific area or equipment with problems in the power grid; Establish an early warning mechanism. When a weak point in operation is detected, the system alarm will notify the relevant operation and maintenance personnel and management personnel, and provide detailed early warning information, including the specific location, type, possible cause and recommended treatment measures of the weak point; the recommended treatment measures include load adjustment, equipment maintenance, line adjustment and modification, and configuration adjustment; Locate the fault point, combine historical data and model prediction results, determine the cause of the fault, and implement emergency repair plans; at the same time, record the fault handling process and results, and update the fault database; formulate a detailed preventive maintenance plan, including inspection and maintenance of key equipment and lines within a fixed period.

6. A power grid operation weak point identification system based on the transmission and distribution network topology structure using the method according to any one of claims 1 to 5, characterized in that: The acquisition unit collects and pre-processes the topological structure data of the power transmission and distribution network, establishes a power grid topology model, and uses graph theory and network simplification technology to simplify the model; An adjustment unit, which reads data quality information from monitoring equipment and sensors, calculates and evaluates the data quality information, and adjusts the data quality information according to the evaluation results; The processing unit extracts key features based on the simplified power grid topology model and the adjusted data quality information, and uses a feature representation method to convert the topology structure information into a feature vector that can be processed by the model; The model optimization unit uses deep learning algorithms to train the model on the extracted feature matrix; it adjusts the model parameters through cross-validation and hyperparameter tuning; The computing unit deploys the trained model in the real-time monitoring system of the power grid and analyzes and processes the real-time collected operation data; it calculates and obtains the operation defects through the prediction results of the model; The output unit identifies the operational weaknesses of the power grid in real time based on the evaluation results, provides early warnings and recommended measures, and performs fault handling and prevention.

7. A computer device comprising: Memory and processor; The memory stores a computer program, wherein the processor implements the steps of any one of the methods of claims 1-5 when executing the computer program.

8. A computer-readable storage medium having a computer program stored thereon, 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 5 are implemented.

Images