An artificial intelligence-based power grid fault analysis method and system

Abstract

The application provides a power grid fault analysis method and system based on artificial intelligence, and relates to the technical field of power grid fault analysis.The method comprises the following steps: collecting historical fault data, performing data clustering analysis on the fault types, and forming historical fault clustering data; according to the historical fault clustering data, performing correlation analysis on the parameter changes of the same type of fault, and obtaining a fault correlation prediction model; obtaining real-time monitoring data, combining the fault correlation prediction model to perform fault prediction analysis, and forming real-time fault prediction results.The method can realize reasonable prediction of faults through reasonable data analysis, and provide more accurate and effective detailed fault parameters for fault processing.

Description

Technical Field

[0001] This invention relates to the field of power grid fault analysis technology, and more specifically, to a power grid fault analysis method and system based on artificial intelligence. Background Technology

[0002] Fault analysis is a crucial part of power grid maintenance. It provides accurate information such as operating methods, locations, and timing for subsequent maintenance. Accurate fault analysis results significantly improve the efficiency of fault handling. After all, the power grid is a massive system, and manual fault diagnosis consumes a great deal of time and manpower, and accurate judgment is not always possible. Furthermore, fault analysis requires processing a large amount of data, placing immense pressure on manual troubleshooting.

[0003] Currently, big data-based analysis methods are also used. While this approach achieves efficient processing of large amounts of data, it lacks predictive capabilities. It only extracts features from historically reproducible faults to make judgments and cannot predict all newly occurring faults, especially those of the same type but with different handling methods.

[0004] Therefore, designing an artificial intelligence-based power grid fault analysis method and system that can make reasonable fault predictions through reasonable data analysis and provide more accurate and effective detailed fault parameters for fault handling is an urgent problem to be solved. Summary of the Invention

[0005] The purpose of this invention is to provide an artificial intelligence-based power grid fault analysis method. This method involves the rational and effective clustering of historical fault data to form basic analytical data for different fault types. This data is then used as training data for building an intelligent model. Predictive model analysis is performed on the fault cluster data for each fault type, establishing the correlation between the changes in selected baseline fault parameters and the changes in selected monitoring parameters. This allows for the accurate and effective determination of the specific fault parameters corresponding to the fault, based solely on the changes in monitoring parameters relative to the baseline monitoring parameters. This provides reliable, accurate, and detailed fault parameter information for subsequent targeted fault handling. Compared to relying solely on historical big data for fault type judgment, the artificial intelligence prediction model, by deeply analyzing the patterns of fault parameter changes, can predict specific fault parameter information, significantly improving the accuracy of fault analysis and providing more detailed reference information for fault handling, thus facilitating accurate and efficient fault handling.

[0006] The present invention also aims to provide an artificial intelligence-based power grid fault analysis system. This system acquires data through a data acquisition unit, performs reasonable data cleaning and clustering of historical big data in a clustering processing unit, providing effective and accurate basic training data for subsequent artificial intelligence training. A training prediction unit processes, analyzes, and learns from this basic training data, forming a predictive fault analysis model. A real-time analysis unit then uses this model to accurately predict fault parameter values ​​using only real-time monitoring data. The close integration of different functional units forms an artificial intelligence predictive analysis model for power grid faults, providing a crucial material foundation for intelligent fault prediction and analysis.

[0007] In a first aspect, the present invention provides an artificial intelligence-based power grid fault analysis method, comprising: collecting historical fault data, performing data clustering analysis for fault types to form historical fault cluster data; performing correlation analysis on parameter changes of the same type of fault based on the historical fault cluster data to obtain a fault correlation prediction model; and acquiring real-time monitoring data, combining the fault correlation prediction model to perform fault prediction analysis to form a real-time fault prediction result.

[0008] In this invention, the method forms basic analysis data for different fault types through reasonable and effective clustering of historical fault data. This data is then used as training data for building an intelligent model. Predictive model analysis is performed on the fault cluster data for each fault type, establishing the correlation between the change information of the selected baseline fault parameter data and the change information of the selected monitoring parameter data. Thus, based solely on the change of the monitoring parameters relative to the baseline monitoring parameter data, the predictive model can accurately and effectively determine the specific fault parameters corresponding to the fault, providing reliable, accurate, and detailed fault parameter information for subsequent targeted fault handling. Compared to judging the fault type using only historical big data, the artificial intelligence predictive model can predict the specific parameter information of the fault based on a deep understanding of the change patterns of fault parameters, greatly improving the accuracy of fault analysis and providing more detailed reference information for fault handling, which is conducive to the accurate and efficient implementation of fault handling.

[0009] One possible approach is to collect historical fault data, perform data clustering analysis based on fault type, and form historical fault cluster data. This includes: extracting fault parameter information and corresponding monitoring parameter information for each fault occurrence from the historical fault data to form single historical fault information; clustering different single historical fault information according to the corresponding fault clusters to form different historical fault type cluster sets; and combining different historical fault type cluster sets to form historical fault cluster data.

[0010] In this invention, clustering historical fault data by fault type primarily provides more effective and accurate foundational data for training predictive models for different types of faults. Clustering analysis uses data from a single fault occurrence as the basic unit for clustering. Fault data includes monitoring parameters collected at the time of the fault and corresponding fault information. Monitoring parameters mainly include operational parameters such as voltage, impedance, and current, and environmental parameters such as temperature and humidity. Fault information includes parameters that measure the magnitude of the fault, such as location and duration of damage for line segments, and lightning strike range, depth, and intensity for insulators. It is important to note that since reasonable model prediction is required for the same fault clustering data, the monitoring parameter data and fault parameter data must ensure that the types of these parameters are the same across different historical fault instances for the same fault type. This provides a consistent data foundation for subsequent predictive analysis to determine the patterns of parameter changes based on fault information. It is understandable that for fault parameters and monitoring parameters, even for the same type of fault, there will be differences in parameter data. Therefore, the established prediction model explores the patterns of these differences in parameter data, learns from the training of basic data to obtain an accurate representation of these patterns, and thus makes accurate and effective predictions even for fault parameters that have never occurred in historical data.

[0011] One possible approach is to perform correlation analysis on parameter changes for the same type of fault based on historical fault clustering data to obtain a fault correlation prediction model. This includes: digitizing parameter information for different historical fault type clusters in the historical fault clustering data to form digitized single-instance historical fault information corresponding to different single-instance historical fault information within the historical fault type clusters; performing baseline parameter analysis based on parameter correlation based on the digitized single-instance historical fault information corresponding to different single-instance historical fault information within the historical fault type clusters to determine the fault type baseline parameter information corresponding to the historical fault type clusters; and performing parameter change correlation analysis relative to the baseline parameters based on the fault type baseline parameter information, combined with the digitized single-instance historical fault information under the historical fault type clusters, to establish a fault correlation prediction model corresponding to the fault type.

[0012] In this invention, a predictive model is established by combining correlation analysis with historical fault clustering data. The aim is to use historical clustering data for training to uncover the mapping relationship between the change in fault parameters relative to a reasonable fault parameter and the change in monitoring parameters relative to a reasonable monitoring parameter. This allows for accurate and effective prediction of fault parameters during subsequent real-time monitoring by acquiring information on the changes in monitoring parameters relative to the benchmark values. The predictive model can then determine the changes in the corresponding fault parameters relative to the benchmark fault parameters, thereby predicting specific fault parameter values. This eliminates the need for on-site or more complex analyses to obtain specific fault parameter values. To establish this predictive model, it is first necessary to find reasonable fault parameter and monitoring parameter benchmarks for the clustered data. Here, correlation analysis can be used to select the fault and monitoring information with the strongest correlation as basic reference data. Secondly, based on the basic reference information, the changes in different fault and monitoring data within the clustered data relative to the benchmark reference data are determined. Finally, the mapping relationship between these monitoring changes and fault changes is used to analyze and establish the predictive model, forming a reasonable and accurate predictive model. It's important to note that the mapping relationship between changes in fault parameters and changes in monitoring parameters exhibits a non-linearity due to the varying nature of these parameter changes. Therefore, big data analysis can be used to learn this non-linear pattern and fully grasp the changing patterns of the mapping relationship between fault parameter changes and monitoring parameter changes. When conducting correlation analysis between changes in fault parameters and monitoring parameter changes, it's crucial to ensure the consistency of the matrix elements.

[0013] As one possible implementation, parameter information is digitized from different historical fault type clusters in the historical fault clustering data to form digitized single historical fault data corresponding to different single historical fault information in the historical fault type clusters. This includes: extracting fault parameter values ​​corresponding to different fault parameters for different single historical fault information in different historical fault type clusters, and establishing a single historical fault parameter matrix corresponding to single historical fault information. Where m represents the number of different single historical fault information in the historical fault type cluster, and n represents the location number of different fault parameters in the matrix; for different single historical fault information in different historical fault type clusters, the monitoring parameter values ​​of the corresponding different monitoring parameters are extracted, and a single historical monitoring parameter matrix corresponding to the single historical fault information is established. Where k represents the location number of different monitoring parameters in the matrix; for different single historical fault information in different historical fault type clusters, combined with the corresponding single historical fault parameter matrix. and single historical monitoring parameter matrix This generates corresponding historical fault data.

[0014] In this invention, matrixing the monitoring and fault parameter values ​​acquired at each fault occurrence facilitates rational and effective data processing, enabling the organization and effective extraction of training data. Of course, during matrixing, it is essential to ensure that the positions of different parameters in the monitoring and fault parameter matrices corresponding to different faults are fixed. This guarantees the rationality and effectiveness of data processing during subsequent training and analysis using the matrix data. The type of matrix established can be set according to needs, as long as it facilitates subsequent analysis. Furthermore, all parameter values ​​can be directly obtained raw values ​​or indirect data obtained through calculations, but it is necessary to ensure that the parameter values ​​have a unified data benchmark, including units and reference standards, to effectively guarantee the rationality of the data.

[0015] As one possible implementation, based on the digitized information of different single historical faults in the historical fault type clusters, a baseline parameter analysis based on parameter correlation is performed to determine the fault type baseline parameter information corresponding to the historical fault type clusters. This includes: combining different digitized information of different single historical faults in the single historical fault information with different single historical fault parameter matrices. This generates a set of historical fault type fault parameter matrices; based on this set, matrix correlation analysis is performed to determine the basic fault parameter matrix. Based on the information from a single historical fault, the basic fault parameter matrix is ​​determined. Corresponding single historical monitoring parameter matrix And calibrate the basic monitoring parameter matrix. Combined with the basic fault parameter matrix and the corresponding basic monitoring parameter matrix This generates baseline parameter information for the fault type corresponding to the historical fault type cluster set.

[0016] In this invention, the selected baseline fault parameter matrix is ​​the fault matrix that has the closest correlation with other fault parameter matrices in historical fault data of the same type. After all, the selection of the baseline parameter matrix has a significant impact on the accuracy of subsequent prediction models. Using the most closely correlated matrix as the baseline parameter matrix ensures that the basic data for subsequent analysis has higher accuracy and rationality compared to other historical data. Of course, to ensure that the baseline parameter matrix has high reference value not only as the referenced baseline fault parameter matrix but also as the referenced baseline monitoring parameter matrix, the selection of the baseline monitoring parameters should ensure that their correspondence with the fault parameter matrix accurately and reasonably reflects the relationship between the two. Therefore, after obtaining the baseline fault parameter matrix, the monitoring parameter matrix is ​​directly determined based on the monitoring parameter matrix corresponding to the baseline fault parameter matrix in historical fault type data.

[0017] As one possible implementation, a matrix correlation analysis is performed based on the historical fault type fault parameter matrix set to determine the basic fault parameter matrix. This includes: different single historical fault parameter matrices in the historical fault type fault parameter matrix set. The parameter matrix for each individual historical fault is determined according to the following method. Corresponding correlation measure For any single historical fault parameter matrix Determine the parameter matrix of each other single historical fault in the fault parameter matrix set of the same historical fault type. The cosine similarity is calculated; based on all the determined cosine similarities, the sum of the cosine similarities is calculated and calibrated as the relevance measure. Compare the parameter matrices of different single historical faults. correlation measure and the highest correlation measure Corresponding single historical fault parameter matrix The basic fault parameter matrix corresponding to the historical fault type cluster set is determined. .

[0018] In this invention, during the determination of the basic fault parameter matrix, the correlation between parameter data exists between different occurrences of the same type of fault. This correlation reflects the inherent regularity of parameters in the same type of fault. Furthermore, the occurrence of fault parameter values ​​is probabilistic. If the selection of the benchmark fault parameter matrix ensures that it represents a combination of parameter values ​​with a high frequency of occurrence for that type of fault, then using it as a benchmark for regular analysis of parameter value changes can improve the rationality and accuracy of the analysis results. Therefore, utilizing matrix correlation, selecting the matrix with the largest sum of correlations with other fault parameter matrices as the benchmark is the most reasonable approach.

[0019] As one possible implementation, based on the fault type baseline parameter information and combined with the digitized information of different single historical faults under the historical fault type cluster set, a correlation analysis of parameter changes relative to the baseline parameters is performed to establish a fault correlation prediction model corresponding to the fault type. This includes: determining the single historical fault parameter matrix corresponding to the digitized information of other single historical faults under the historical fault type cluster set based on the fault type baseline parameter information. The remaining term historical fault parameter matrix is ​​calibrated. 'u' represents the number of the remaining individual historical fault data after the baseline parameter information of the fault type is removed from the historical fault type clustering set; for different remaining terms of the historical fault parameter matrix... Determine the relative basic fault parameter matrix Relative fault parameter change matrix ,in, Based on the fault type baseline parameter information, determine the single historical monitoring parameter matrix corresponding to other single historical fault data under the historical fault type cluster set. The remaining term is defined as the historical monitoring parameter matrix. ; For different residual terms historical monitoring parameter matrices Determine the relative basic monitoring parameter matrix relative monitoring parameter change matrix ,in, ; The relative fault parameter change matrix corresponding to the same single historical fault data. and relative monitoring parameter change matrix By combining these data, we can generate different relative parameter change information for each individual fault; by combining these different relative parameter change information for each individual fault, we can determine the maximum relative monitoring parameter value for each different monitoring parameter. and the minimum value of relative monitoring parameters This forms the relative variation range of the corresponding fault monitoring parameters. Collect information on relative parameter changes from different single faults, perform correlation analysis on parameter changes, and combine this with the relative change range of fault monitoring parameters corresponding to different monitoring parameters. Establish a fault correlation prediction model corresponding to the fault type.

[0020] In this invention, the fault correlation prediction model is established based on the relationship between the changes in fault parameters and monitoring parameters relative to their corresponding benchmark values. Therefore, it is first necessary to determine the changes in fault parameters relative to benchmark fault parameters and the changes in monitoring parameters relative to benchmark monitoring parameters based on historical fault data. Secondly, the large dataset used for model training is extensive, so the range of parameter changes can be measured using historical data. This allows us to obtain possible combinations of monitoring parameter changes when this type of fault occurs, providing a reference for fault type prediction in real-time monitoring. Of course, performing correlation analysis between the changes in fault parameters and monitoring parameters based on the obtained changes is a crucial foundation for establishing the prediction model. This model fully demonstrates the correlation between changes in fault parameters and changes in monitoring parameters, providing a reasonable and accurate analytical model for predicting fault parameters after obtaining monitoring parameters.

[0021] One possible approach is to aggregate relative parameter change information from different single faults, perform correlation analysis on parameter changes, and combine this with the relative change range of fault monitoring parameters corresponding to different monitoring parameters. Establish a fault correlation prediction model corresponding to fault type, including: generating a relative fault parameter change matrix for different single fault relative parameter change information. and relative monitoring parameter change matrix The matrix elements are uniformly processed to form the corresponding relative fault uniform parameter change matrix. and relative monitoring unified parameter change matrix Where the maximum value of i is the largest of the maximum values ​​of k and n; a unified correlation prediction formula is established for the relative parameter changes of different single faults. : Where v represents the number of different fault types, This represents the relative fault prediction parameter matrix. This represents the matrix representing the relative real-time monitoring parameter changes. This represents the transpose of the matrix representing the changes in the relative real-time monitoring parameters. This represents a matrix of power-law correlation parameters. This represents the single-power correlation parameter matrix. Represents a constant parameter matrix; for different single-fault relative parameter change information, the corresponding relative fault unified parameter change matrix is ​​used. As a correlation prediction formula The relative fault prediction parameter matrix in the matrix will be compared with the relative monitoring unified parameter change matrix. As a correlation prediction formula The relative real-time monitoring parameter change matrix is ​​used to solve for the multi-power correlation parameter matrix, the single-power correlation parameter matrix, and the constant parameter matrix; the correlation prediction formula after solving is then used. The relative variation range of fault monitoring parameters and corresponding different monitoring parameters The fault correlation prediction model corresponding to the fault type was determined.

[0022] In this invention, the establishment of the fault correlation prediction model mainly involves determining the mapping relationship between the relative parameter change information of the fault and the monitoring parameter change information. This mapping relationship is not a simple independent mapping between individual parameters, but rather an overall correlation mapping established through matrix data. Considering that this mapping relationship is non-linear, the provided correlation prediction formula is also multi-dimensional. It includes establishing the correlation relationship with the fault parameter matrix using the inner product of the monitoring parameter matrix and the monitoring parameter matrix itself. The corresponding power parameter matrix and constant parameter matrix are the parameters that need to be solved for formula formulation. Using the relative parameter change information of a single fault, the data volume corresponding to the total number of unknowns in the formula is obtained for solution, thereby determining the specific parameter value of each element in these power parameter matrices and constant parameter matrices. Of course, the solution method is diverse, and it can be achieved by using simultaneous equations. Solving the problem can also be done by combining the characteristics of the matrix with assumptions, eigenvalues, etc. Of course, before solving the problem, it is more important that the matrix in the established formula is computable. Therefore, before establishing the formula, it is necessary to perform reasonable element unification processing on the monitoring parameter matrix and the fault parameter matrix. That is, while keeping the positions of the original matrix elements unchanged, a unified square matrix is ​​established based on the maximum number of elements in the two matrices. In this way, the inner product and addition and subtraction operations in the formula can be performed smoothly. For the extra elements generated in the process of establishing the unified square matrix, it is acceptable to use unit quantities to replace them, as long as it does not affect the accuracy of the calculation results.

[0023] One possible approach is to acquire real-time monitoring data, combine it with a fault correlation prediction model to perform fault prediction analysis, and generate real-time fault prediction results. This includes: based on real-time monitoring data, combined with the basic monitoring parameter matrix corresponding to different fault types. The corresponding real-time monitoring parameter change matrix is ​​determined; the relative change range of fault monitoring parameters is determined based on different fault types. It was determined that the changes in all monitoring parameters in the real-time monitoring parameter change matrix belong to the relative change range of the corresponding fault monitoring parameters under the fault type. The fault types are identified and labeled as real-time predicted fault types; the fault correlation prediction model corresponding to the real-time predicted fault type is obtained, and combined with the real-time monitoring parameter change matrix, the real-time fault parameter change information is determined; based on the real-time fault parameter change information, combined with the basic fault parameter matrix corresponding to the real-time predicted fault type, the fault type is further analyzed. The parameter values ​​of different fault parameters are determined and real-time fault prediction results are generated.

[0024] In this invention, real-time fault prediction analysis first requires identifying possible fault types, and then using the prediction model corresponding to each fault type to perform predictive analysis and determine specific fault parameter values. Here, the determination of fault types is mainly based on the range of changes in monitoring parameters corresponding to different fault types. It can be understood that only when the changes in all monitoring parameters fall within their corresponding ranges can it be proven that the current monitoring parameters have indeed experienced a fault and have a definite fault type. Therefore, the change in fault parameters is determined using the prediction model for the identified fault type, and then combined with the baseline fault parameters corresponding to the fault type, the fault parameter values ​​at the time of the real-time fault occurrence are determined. It should be noted that when using the prediction model to analyze the changes in fault parameters, the monitoring parameters still need to be standardized to ensure the analyzability and computability of the predictive analysis model.

[0025] Secondly, the present invention provides an artificial intelligence-based power grid fault analysis system, comprising: a data acquisition unit for acquiring historical fault data and real-time monitoring data; a clustering processing unit for performing data clustering analysis on the historical fault data acquired by the data acquisition unit according to fault type to form historical fault cluster data; a training and prediction unit for performing correlation analysis on the changes of parameters of the same type of fault based on the historical fault cluster data formed by the clustering processing unit to obtain a fault correlation prediction model; and a real-time analysis unit for performing fault prediction analysis based on the real-time monitoring data acquired by the data acquisition unit and combined with the fault correlation prediction model formed by the training and prediction unit to form a real-time fault prediction result.

[0026] In this invention, the system acquires data through a data acquisition unit, performs reasonable data cleaning and clustering of historical big data in a clustering processing unit, providing effective and accurate basic training data for subsequent artificial intelligence training. A training prediction unit processes, analyzes, and learns from this basic training data, forming a predictive fault analysis model. The real-time analysis unit then uses this fault analysis prediction model to accurately predict fault parameter values ​​using only real-time monitoring data. The close integration of different functional units forms an artificial intelligence predictive analysis model for power grid faults, providing a crucial material foundation for intelligent fault prediction and analysis.

[0027] The beneficial effects of the power grid fault analysis method and system based on artificial intelligence provided by this invention are as follows: This method generates basic analytical data for different fault types through reasonable and effective clustering of historical fault data. This data is then used as training data for building an intelligent model. Predictive model analysis is performed on the fault cluster data for each fault type, establishing the correlation between the changes in the selected baseline fault parameters and the changes in the selected monitoring parameters. This allows for the accurate and effective determination of the specific fault parameters corresponding to the fault, based solely on the changes in the monitoring parameters relative to the baseline monitoring parameters. This provides reliable, accurate, and detailed fault parameter information for subsequent targeted fault handling. Compared to relying solely on historical big data for fault type judgment, the artificial intelligence prediction model, by deeply analyzing the patterns of fault parameter changes, can predict specific fault parameter information, significantly improving the accuracy of fault analysis and providing more detailed reference information for fault handling, thus facilitating accurate and efficient implementation of fault handling.

[0028] The system collects data through a data acquisition unit, performs reasonable data cleaning and clustering of historical big data in the clustering processing unit, providing effective and accurate basic training data for subsequent artificial intelligence training. The training and prediction unit processes, analyzes, and learns from this basic training data, forming a predictive fault analysis model. The real-time analysis unit then uses this model to accurately predict fault parameter values ​​using only real-time monitoring data. The close integration of different functional units forms an artificial intelligence predictive analysis model for power grid faults, providing a crucial material foundation for intelligent fault prediction and analysis. Attached Figure Description

[0029] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 A flowchart illustrating the steps of an artificial intelligence-based power grid fault analysis method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of an artificial intelligence-based power grid fault analysis system provided in an embodiment of the present invention. Detailed Implementation

[0031] The technical solutions of the present invention will now be described with reference to the accompanying drawings in the embodiments of the present invention.

[0032] Fault analysis is a crucial part of power grid maintenance. It provides accurate information such as operating methods, locations, and timing for subsequent maintenance. Accurate fault analysis results significantly improve the efficiency of fault handling. After all, the power grid is a massive system, and manual fault diagnosis consumes a great deal of time and manpower, and accurate judgment is not always possible. Furthermore, fault analysis requires processing a large amount of data, placing immense pressure on manual troubleshooting.

[0033] Currently, big data-based analysis methods are also used. While this approach achieves efficient processing of large amounts of data, it lacks predictive capabilities. It only extracts features from historically reproducible faults to make judgments and cannot predict all newly occurring faults, especially those of the same type but with different handling methods.

[0034] refer to Figures 1-2This invention provides an artificial intelligence-based power grid fault analysis method. This method forms basic analysis data for different fault types through reasonable and effective clustering of historical fault data. This data is then used as training data for building an intelligent model. Predictive model analysis is performed on the fault cluster data for each fault type, establishing the correlation between the changes in selected baseline fault parameters and the changes in selected monitoring parameters. Therefore, based solely on the changes in monitoring parameters relative to the baseline monitoring parameters, the predictive model can accurately and effectively determine the specific fault parameters corresponding to the fault. This provides reliable, accurate, and detailed fault parameter information for subsequent targeted fault handling. Compared to judging fault types solely using historical big data, the artificial intelligence predictive model, based on in-depth analysis of fault parameter change patterns, can predict specific fault parameter information, greatly improving the accuracy of fault analysis and providing more detailed reference information for fault handling, thus facilitating accurate and efficient fault handling.

[0035] An artificial intelligence-based power grid fault analysis method specifically includes the following steps: S1: Collect historical fault data, perform data clustering analysis based on fault type, and form historical fault cluster data.

[0036] Collect historical fault data and perform data clustering analysis based on fault type to form historical fault cluster data. This includes: extracting fault parameter information and corresponding monitoring parameter information for each fault occurrence from historical fault data to form single historical fault information; clustering different single historical fault information according to the corresponding fault cluster to form different historical fault type cluster sets; and combining different historical fault type cluster sets to form historical fault cluster data.

[0037] 47. Clustering historical fault data by fault type primarily provides a more effective and accurate foundation for training predictive models for different types of faults. Cluster analysis uses data from a single fault occurrence as the basic unit for clustering. Fault data includes monitoring parameters collected at the time of the fault and corresponding fault information. Monitoring parameters mainly include operational parameters such as voltage, impedance, and current, and environmental parameters such as temperature and humidity. Fault information includes parameters that measure the magnitude of the fault, such as location and duration of line segments, and lightning strike range, depth, and intensity of damage to insulators. It is important to note that since reasonable model predictions are needed for clustered data of the same fault, the types of monitoring parameters and fault parameters must be consistent across different historical fault instances for the same fault type. This ensures a unified data foundation for subsequent predictive analysis to determine the patterns of parameter changes based on fault information. It is understandable that for fault parameters and monitoring parameters, even for the same type of fault, there will be differences in parameter data. Therefore, the established prediction model explores the patterns of these differences in parameter data, learns from the training of basic data to obtain an accurate representation of these patterns, and thus makes accurate and effective predictions even for fault parameters that have never occurred in historical data.

[0038] S2: Based on historical fault clustering data, perform correlation analysis on the changes in parameters of the same type of fault to obtain a fault correlation prediction model.

[0039] Based on historical fault clustering data, correlation analysis of parameter changes for the same type of fault is conducted to obtain a fault correlation prediction model. This includes: digitizing parameter information for different historical fault type clusters in the historical fault clustering data to form digitized single-instance historical fault information corresponding to different single-instance historical fault information within the historical fault type clusters; performing baseline parameter analysis based on parameter correlation based on the digitized single-instance historical fault information corresponding to different single-instance historical fault information within the historical fault type clusters to determine the fault type baseline parameter information corresponding to the historical fault type clusters; and conducting correlation analysis of parameter changes relative to the baseline parameters based on the fault type baseline parameter information and combined with the digitized single-instance historical fault information under the historical fault type clusters to establish a fault correlation prediction model corresponding to the fault type.

[0040] A predictive model is built by combining correlation analysis with historical fault clustering data. The aim is to use this historical clustering data for training to uncover the mapping relationship between the change in fault parameters relative to a reasonable fault parameter and the change in monitoring parameters relative to a reasonable monitoring parameter. This allows for accurate and effective prediction of fault parameters during subsequent real-time monitoring by acquiring information on the changes in monitoring parameters relative to the benchmark values. The predictive model can then determine the corresponding changes in fault parameters relative to the benchmark fault parameters, thereby predicting specific fault parameter values. This eliminates the need for on-site or more complex analyses to obtain specific fault parameter values. To build this predictive model, it is first necessary to find reasonable fault parameter and monitoring parameter benchmarks for the clustered data. Here, correlation analysis can be used to select the fault and monitoring information with the strongest correlation as basic reference data. Secondly, using this basic reference information as a benchmark, the changes in different fault and monitoring data within the clustered data relative to the benchmark reference data are determined. Finally, the mapping relationship between these monitoring changes and fault changes is used to analyze and build the predictive model, resulting in a reasonable and accurate predictive model. It's important to note that the mapping relationship between changes in fault parameters and changes in monitoring parameters exhibits a non-linearity due to the varying nature of these parameter changes. Therefore, big data analysis can be used to learn this non-linear pattern and fully grasp the changing patterns of the mapping relationship between fault parameter changes and monitoring parameter changes. When conducting correlation analysis between changes in fault parameters and monitoring parameter changes, it's crucial to ensure the consistency of the matrix elements.

[0041] For different historical fault type clusters in the historical fault clustering data, parameter information is digitized to form digitized information of individual historical faults corresponding to different individual historical faults in the historical fault type clusters. This includes: extracting the fault parameter values ​​of different fault parameters for different individual historical faults in different historical fault type clusters, and establishing a single historical fault parameter matrix corresponding to the individual historical fault information. Where m represents the number of different single historical fault information in the historical fault type cluster, and n represents the location number of different fault parameters in the matrix; for different single historical fault information in different historical fault type clusters, the monitoring parameter values ​​of the corresponding different monitoring parameters are extracted, and a single historical monitoring parameter matrix corresponding to the single historical fault information is established. Where k represents the location number of different monitoring parameters in the matrix; for different single historical fault information in different historical fault type clusters, combined with the corresponding single historical fault parameter matrix. and single historical monitoring parameter matrix This generates corresponding historical fault data.

[0042] Matrixing the monitoring and fault parameter values ​​acquired at each fault occurrence facilitates reasonable and effective data processing, enabling the organization and effective extraction of training data. However, during matrixing, it's crucial to ensure that the positions of different parameters within the monitoring and fault parameter matrices corresponding to different faults remain fixed. This guarantees the rationality and effectiveness of data processing during subsequent training and analysis using the matrix data. The type of matrix can be set according to needs, facilitating subsequent analysis. Furthermore, all parameter values ​​can be directly obtained raw values ​​or indirect data obtained through calculations, but a unified data benchmark is essential, including units and reference standards, to ensure data validity.

[0043] Based on the digitized data of single historical faults corresponding to different single historical fault information in the historical fault type clusters, benchmark parameter analysis based on parameter correlation is performed to determine the fault type benchmark parameter information corresponding to the historical fault type clusters. This includes: combining different digitized data of single historical faults with different single historical fault parameter matrices. This generates a set of historical fault type fault parameter matrices; based on this set, matrix correlation analysis is performed to determine the basic fault parameter matrix. Based on the information from a single historical fault, the basic fault parameter matrix is ​​determined. Corresponding single historical monitoring parameter matrix And calibrate the basic monitoring parameter matrix. Combined with the basic fault parameter matrix and the corresponding basic monitoring parameter matrix This generates baseline parameter information for the fault type corresponding to the historical fault type cluster set.

[0044] The baseline fault parameter matrix selected here is the fault matrix that has the closest correlation with other fault parameter matrices in historical fault data of the same type. After all, the selection of the baseline parameter matrix has a significant impact on the accuracy of subsequent prediction models, and using the most closely correlated matrix as the baseline parameter matrix ensures that the basic data for subsequent analysis has a higher accuracy and rationality compared to other historical data. Of course, to ensure that the baseline parameter matrix has high reference value not only in the referenced baseline fault parameter matrix, but also in the referenced baseline monitoring parameter matrix, the selection of the baseline monitoring parameters should ensure that their correspondence with the fault parameter matrix accurately and reasonably reflects the relationship between the two. Therefore, after obtaining the baseline fault parameter matrix, the monitoring parameter matrix is ​​directly determined based on the monitoring parameter matrix corresponding to the baseline fault parameter matrix in the historical fault type data.

[0045] Based on the historical fault parameter matrix set, matrix correlation analysis is performed to determine the basic fault parameter matrix. This includes: different single historical fault parameter matrices in the historical fault type fault parameter matrix set. The parameter matrix for each individual historical fault is determined according to the following method. Corresponding correlation measure For any single historical fault parameter matrix Determine the parameter matrix of each other single historical fault in the fault parameter matrix set of the same historical fault type. The cosine similarity is calculated; based on all the determined cosine similarities, the sum of the cosine similarities is calculated and calibrated as the relevance measure. Compare the parameter matrices of different single historical faults. correlation measure and the highest correlation measure Corresponding single historical fault parameter matrix The basic fault parameter matrix corresponding to the historical fault type cluster set is determined. .

[0046] In determining the basic fault parameter matrix, the correlation between parameter data exists between different occurrences of the same type of fault. This correlation reflects the inherent regularity of parameters in the same type of fault. Furthermore, the occurrence of fault parameter values ​​is probabilistic. If the selection of the benchmark fault parameter matrix ensures that it represents a combination of parameter values ​​with high frequency of occurrence for that type of fault, then using it as a benchmark for analyzing the regularity of parameter value changes can improve the rationality and accuracy of the analysis results. Therefore, utilizing matrix correlation, selecting the matrix with the largest sum of correlations with other fault parameter matrices as the benchmark is the most reasonable approach.

[0047] Based on the baseline parameter information of the fault type, and combined with the digitized information of different single historical faults under the historical fault type cluster set, a correlation analysis of parameter changes relative to the baseline parameter is performed to establish a fault correlation prediction model corresponding to the fault type. This includes: determining the single historical fault parameter matrix corresponding to the digitized information of other single historical faults under the historical fault type cluster set based on the baseline parameter information of the fault type. The remaining term historical fault parameter matrix is ​​calibrated. 'u' represents the number of the remaining individual historical fault data after the baseline parameter information of the fault type is removed from the historical fault type clustering set; for different remaining terms of the historical fault parameter matrix... Determine the relative basic fault parameter matrix Relative fault parameter change matrix ,in, Based on the fault type baseline parameter information, determine the single historical monitoring parameter matrix corresponding to other single historical fault data under the historical fault type cluster set. The remaining term is defined as the historical monitoring parameter matrix. ; For different residual terms historical monitoring parameter matrices Determine the relative basic monitoring parameter matrix relative monitoring parameter change matrix ,in, ; The relative fault parameter change matrix corresponding to the same single historical fault data. and relative monitoring parameter change matrix By combining these data, we can generate different relative parameter change information for each individual fault; by combining these different relative parameter change information for each individual fault, we can determine the maximum relative monitoring parameter value for each different monitoring parameter. and the minimum value of relative monitoring parameters This forms the relative variation range of the corresponding fault monitoring parameters. Collect information on relative parameter changes from different single faults, perform correlation analysis on parameter changes, and combine this with the relative change range of fault monitoring parameters corresponding to different monitoring parameters. Establish a fault correlation prediction model corresponding to the fault type.

[0048] The establishment of a fault correlation prediction model focuses on the relationship between the changes in fault parameters and monitoring parameters relative to their corresponding baseline values. Therefore, it is first necessary to determine the changes in fault parameters relative to baseline fault parameters and the changes in monitoring parameters relative to baseline monitoring parameters based on historical fault data. Secondly, the large dataset used for model training is extensive, allowing the range of parameter changes to be measured using historical data. This enables the identification of possible combinations of monitoring parameter changes when this type of fault occurs, providing a reference for fault type prediction in real-time monitoring. Of course, conducting correlation analysis between the changes in fault parameters and monitoring parameters based on the obtained changes is a crucial foundation for establishing the prediction model. This model fully demonstrates the correlation between changes in fault parameters and changes in monitoring parameters, providing a reasonable and accurate analytical model for predicting fault parameters after obtaining the monitoring parameters.

[0049] By aggregating relative parameter change information from different single faults, correlation analysis of parameter changes is performed, and the relative change range of fault monitoring parameters corresponding to different monitoring parameters is considered. Establish a fault correlation prediction model corresponding to fault type, including: generating a relative fault parameter change matrix for different single fault relative parameter change information. and relative monitoring parameter change matrix The matrix elements are uniformly processed to form the corresponding relative fault uniform parameter change matrix. and relative monitoring unified parameter change matrix Where the maximum value of i is the largest of the maximum values ​​of k and n; a unified correlation prediction formula is established for the relative parameter changes of different single faults. : Where v represents the number of different fault types, This represents the relative fault prediction parameter matrix. This represents the matrix representing the relative real-time monitoring parameter changes. This represents the transpose of the matrix representing the changes in the relative real-time monitoring parameters. This represents a matrix of power-law correlation parameters. This represents the single-power correlation parameter matrix. Represents a constant parameter matrix; for different single-fault relative parameter change information, the corresponding relative fault unified parameter change matrix is ​​used. As a correlation prediction formula The relative fault prediction parameter matrix in the matrix will be compared with the relative monitoring unified parameter change matrix. As a correlation prediction formula The relative real-time monitoring parameter change matrix is ​​used to solve for the multi-power correlation parameter matrix, the single-power correlation parameter matrix, and the constant parameter matrix; the correlation prediction formula after solving is then used. The relative variation range of fault monitoring parameters and corresponding different monitoring parameters The fault correlation prediction model corresponding to the fault type was determined.

[0050] The fault correlation prediction model established here primarily determines the mapping relationship between the relative parameter changes in the fault and the changes in the monitoring parameters. This mapping is not a simple independent mapping between individual parameters, but rather an overall correlation mapping established through matrix data. Considering that this mapping relationship is non-linear, the provided correlation prediction formula is also multi-dimensional. It includes using the inner product of the monitoring parameter matrices and the monitoring parameter matrices themselves to establish the correlation with the fault parameter matrix. The corresponding power parameter matrices and constant parameter matrices are the parameters that need to be solved for the formula. Using the relative parameter changes from a single fault, the data volume corresponding to the total number of unknowns in the formula is obtained for solving the problem, thereby determining the specific parameter value of each element in these power parameter matrices and constant parameter matrices. Of course, there are various ways to solve this problem, such as using simultaneous equations. Solving the problem can also be done by combining the characteristics of the matrix with assumptions, eigenvalues, etc. Of course, before solving the problem, it is more important that the matrix in the established formula is computable. Therefore, before establishing the formula, it is necessary to perform reasonable element unification processing on the monitoring parameter matrix and the fault parameter matrix. That is, while keeping the positions of the original matrix elements unchanged, a unified square matrix is ​​established based on the maximum number of elements in the two matrices. In this way, the inner product and addition and subtraction operations in the formula can be performed smoothly. For the extra elements generated in the process of establishing the unified square matrix, it is acceptable to use unit quantities to replace them, as long as it does not affect the accuracy of the calculation results.

[0051] S3: Acquire real-time monitoring data, combine it with a fault correlation prediction model to perform fault prediction analysis, and generate real-time fault prediction results.

[0052] Real-time monitoring data is acquired and combined with a fault correlation prediction model to perform fault prediction analysis, generating real-time fault prediction results, including: based on real-time monitoring data, combined with the basic monitoring parameter matrix corresponding to different fault types. The corresponding real-time monitoring parameter change matrix is ​​determined; the relative change range of fault monitoring parameters is determined based on different fault types. It was determined that the changes in all monitoring parameters in the real-time monitoring parameter change matrix belong to the relative change range of the corresponding fault monitoring parameters under the fault type. The fault types are identified and labeled as real-time predicted fault types; the fault correlation prediction model corresponding to the real-time predicted fault type is obtained, and combined with the real-time monitoring parameter change matrix, the real-time fault parameter change information is determined; based on the real-time fault parameter change information, combined with the basic fault parameter matrix corresponding to the real-time predicted fault type, the fault type is further analyzed. The parameter values ​​of different fault parameters are determined and real-time fault prediction results are generated.

[0053] Real-time fault prediction analysis first requires identifying possible fault types, and then using the prediction model corresponding to each fault type to determine specific fault parameter values. Here, fault type determination is primarily based on the range of changes in monitoring parameters corresponding to different fault types. It can be understood that only when the changes in all monitoring parameters fall within their corresponding ranges can it be proven that a fault has indeed occurred and has a definite fault type. Therefore, using the prediction model for the identified fault type, the changes in fault parameters are determined, and then combined with the baseline fault parameters corresponding to the fault type, the real-time fault parameter values ​​at the time of the fault occurrence are determined. It should be noted that when using the prediction model to analyze the changes in fault parameters, the monitoring parameters still need to be standardized to ensure the analyzability and computability of the prediction analysis model.

[0054] The present invention also provides an artificial intelligence-based power grid fault analysis system, which includes: a data acquisition unit for acquiring historical fault data and real-time monitoring data; a clustering processing unit for performing data clustering analysis on the historical fault data acquired by the data acquisition unit according to fault type to form historical fault cluster data; a training and prediction unit for performing correlation analysis on the changes of parameters of the same type of fault based on the historical fault cluster data formed by the clustering processing unit to obtain a fault correlation prediction model; and a real-time analysis unit for performing fault prediction analysis based on the real-time monitoring data acquired by the data acquisition unit and combined with the fault correlation prediction model formed by the training and prediction unit to form a real-time fault prediction result.

[0055] The system collects data through a data acquisition unit, performs reasonable data cleaning and clustering of historical big data in the clustering processing unit, providing effective and accurate basic training data for subsequent artificial intelligence training. The training and prediction unit processes, analyzes, and learns from this basic training data, forming a predictive fault analysis model. The real-time analysis unit then uses this model to accurately predict fault parameter values ​​using only real-time monitoring data. The close integration of different functional units forms an artificial intelligence predictive analysis model for power grid faults, providing a crucial material foundation for intelligent fault prediction and analysis.

[0056] In summary, the beneficial effects of the artificial intelligence-based power grid fault analysis method and system provided by the embodiments of the present invention are as follows: This method generates basic analytical data for different fault types through reasonable and effective clustering of historical fault data. This data is then used as training data for building an intelligent model. Predictive model analysis is performed on the fault cluster data for each fault type, establishing the correlation between the changes in the selected baseline fault parameters and the changes in the selected monitoring parameters. This allows for the accurate and effective determination of the specific fault parameters corresponding to the fault, based solely on the changes in the monitoring parameters relative to the baseline monitoring parameters. This provides reliable, accurate, and detailed fault parameter information for subsequent targeted fault handling. Compared to relying solely on historical big data for fault type judgment, the artificial intelligence prediction model, by deeply analyzing the patterns of fault parameter changes, can predict specific fault parameter information, significantly improving the accuracy of fault analysis and providing more detailed reference information for fault handling, thus facilitating accurate and efficient implementation of fault handling.

[0057] The system collects data through a data acquisition unit, performs reasonable data cleaning and clustering of historical big data in the clustering processing unit, providing effective and accurate basic training data for subsequent artificial intelligence training. The training and prediction unit processes, analyzes, and learns from this basic training data, forming a predictive fault analysis model. The real-time analysis unit then uses this model to accurately predict fault parameter values ​​using only real-time monitoring data. The close integration of different functional units forms an artificial intelligence predictive analysis model for power grid faults, providing a crucial material foundation for intelligent fault prediction and analysis.

[0058] In the embodiments of this application, "instruction" can include direct and indirect instructions, as well as explicit and implicit instructions. The information indicated by a certain piece of information is called the information to be instructed. In the specific implementation process, there are many ways to instruct the information to be instructed, such as, but not limited to, directly instructing the information to be instructed, such as the information to be instructed itself or its index. It can also indirectly instruct the information to be instructed by instructing other information, where there is a relationship between the other information and the information to be instructed. It can also instruct only a part of the information to be instructed, while the other parts are known or pre-agreed upon. For example, the instruction of specific information can be achieved by using a pre-agreed (e.g., protocol-defined) arrangement of various pieces of information, thereby reducing instruction overhead to some extent. At the same time, common parts of various pieces of information can be identified and uniformly indicated to reduce the instruction overhead caused by individually indicating the same information.

[0059] Furthermore, the specific indication method can also be any existing indication method, such as, but not limited to, the above-mentioned indication methods and their various combinations. Specific details of various indication methods can be found in existing technologies, and will not be repeated here. As described above, for example, when multiple pieces of information of the same type need to be indicated, the indication methods for different pieces of information may differ. In the specific implementation process, the required indication method can be selected according to specific needs. This application embodiment does not limit the selected indication method; therefore, the indication methods involved in this application embodiment should be understood to cover various methods that enable the party to be indicated to obtain the information to be indicated.

[0060] It should be understood that the information to be indicated can be sent as a whole or divided into multiple sub-information messages sent separately, and the sending period and / or timing of these sub-information messages can be the same or different. The specific sending method is not limited in this application embodiment. The sending period and / or timing of these sub-information messages can be predefined, for example, according to a protocol, or configured by the sending device by sending configuration information to the receiving device.

[0061] "Predefined" or "pre-configured" can be achieved by pre-saving corresponding codes, tables, or other means that can be used to indicate relevant information in the device. This application does not limit the specific implementation method. "Saving" can refer to saving in one or more memories. These memories can be separate installations or integrated into the encoder, decoder, processor, or communication device. Alternatively, some memories can be separately installed, while others are integrated into the decoder, processor, or communication device. The type of memory can be any form of storage medium, and this application does not limit this.

[0062] The “protocol” mentioned in the embodiments of this application may refer to a protocol family in the field of communication, a standard protocol with a similar protocol family frame structure, or a related protocol applied to future communication systems. The embodiments of this application do not specifically limit this.

[0063] In the embodiments of this application, descriptions such as "when," "under the circumstances," "if," and "if" all refer to the device making corresponding processing under certain objective circumstances, and are not limited to a specific time. They do not require the device to make a judgment action during implementation, nor do they imply any other limitations.

[0064] In the description of the embodiments of this application, unless otherwise stated, " / " indicates that the objects before and after are in an "or" relationship. For example, A / B can represent A or B. "And / or" in the embodiments of this application is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone, where A and B can be singular or plural. Furthermore, in the description of the embodiments of this application, unless otherwise stated, "multiple" refers to two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple. Additionally, to facilitate a clear description of the technical solutions of the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first," "second," etc., do not limit the quantity or order of execution, and that "first," "second," etc., are not necessarily different. Furthermore, in the embodiments of this application, words such as "exemplary" or "for example" are used to indicate that something is being used as an example, illustration, or description. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner for ease of understanding.

[0065] It should be understood that the processor in the embodiments of this application can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0066] It should also be understood that the memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0067] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0068] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0069] In this application, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0070] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0071] Those skilled in the art will recognize that the units and algorithm steps of the various examples 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 implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art 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.

[0072] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0073] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0074] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0075] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0076] If the aforementioned functions are implemented as 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 this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0077] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A power grid fault analysis method based on artificial intelligence, characterized in that, include: Collect historical fault data, perform data clustering analysis based on fault type, and form historical fault cluster data; Based on the historical fault clustering data, a correlation analysis of the changes in parameters of the same type of fault is performed to obtain a fault correlation prediction model. Real-time monitoring data is acquired and combined with the fault correlation prediction model to perform fault prediction analysis and generate real-time fault prediction results.

2. The power grid fault analysis method based on artificial intelligence according to claim 1, characterized in that, The process involves collecting historical fault data, performing data clustering analysis based on fault type, and forming historical fault cluster data, including: For the historical fault data, extract the fault parameter information and corresponding monitoring parameter information for each fault to form single historical fault information; For different single historical fault information, cluster them according to the corresponding fault cluster to form different historical fault type cluster sets; Different clusters of the aforementioned historical fault types are combined to form the historical fault cluster data.

3. The power grid fault analysis method based on artificial intelligence according to claim 2, characterized in that, The step of performing correlation analysis on the changes in parameters of similar types of faults based on the historical fault clustering data to obtain a fault correlation prediction model includes: For different historical fault type clusters in the historical fault clustering data, the parameter information is digitized to form single historical fault data information corresponding to different single historical fault information in the historical fault type clusters. Based on the single historical fault data information corresponding to different single historical fault information in the historical fault type cluster, a baseline parameter analysis based on parameter correlation is performed to determine the fault type baseline parameter information corresponding to the historical fault type cluster. Based on the fault type baseline parameter information and combined with the different single historical fault data information under the historical fault type cluster, a correlation analysis of parameter changes relative to the baseline parameters is performed to establish the fault correlation prediction model corresponding to the fault type.

4. The power grid fault analysis method based on artificial intelligence according to claim 3, characterized in that, The step of digitizing parameter information for different historical fault type clusters in the historical fault clustering data to form single historical fault data corresponding to different single historical fault information in the historical fault type clusters includes: For different single historical fault information in different clusters of different historical fault types, extract the corresponding fault parameter values ​​and establish a single historical fault parameter matrix corresponding to the single historical fault information. Where m represents the number of different single historical fault information in the historical fault type cluster set, and n represents the location number of different fault parameters in the matrix; For different single historical fault information in different clusters of different historical fault types, extract the corresponding monitoring parameter values ​​of different monitoring parameters, and establish a single historical monitoring parameter matrix corresponding to the single historical fault information. , where k represents the location number of different monitoring parameters in the matrix; For different single historical fault information in different clusters of different historical fault types, combine the corresponding single historical fault parameter matrix. and the single historical monitoring parameter matrix This forms the corresponding digitized information of the single historical fault.

5. The power grid fault analysis method based on artificial intelligence according to claim 4, characterized in that, The step of performing baseline parameter analysis based on parameter correlation based on the single historical fault data corresponding to different single historical fault information in the historical fault type cluster to determine the fault type baseline parameter information corresponding to the historical fault type cluster includes: For different single historical fault data in the single historical fault information, combine different single historical fault parameter matrices This forms a set of historical fault type fault parameter matrices; Based on the historical fault type fault parameter matrix set, matrix correlation analysis is performed to determine the basic fault parameter matrix. ; Based on the single historical fault information, the basic fault parameter matrix is ​​determined. The corresponding single historical monitoring parameter matrix And calibrate the basic monitoring parameter matrix. ; Combining the aforementioned basic fault parameter matrix and the corresponding basic monitoring parameter matrix This forms the fault type baseline parameter information corresponding to the historical fault type cluster set.

6. The power grid fault analysis method based on artificial intelligence according to claim 5, characterized in that, The basic fault parameter matrix is ​​determined by performing matrix correlation analysis based on the historical fault type fault parameter matrix set. ,include: For different single historical fault parameter matrices in the historical fault type fault parameter matrix set Each of the single historical fault parameter matrices is determined according to the following method. Corresponding correlation measure : For any of the single historical fault parameter matrices Determine each of the other single historical fault parameter matrices in the historical fault parameter matrix set that is the same as the historical fault type. Cosine similarity; Based on all the determined cosine similarities, the sum of cosine similarities is calculated and calibrated as the correlation measure. ; Comparison of different single-instance historical fault parameter matrices The correlation measure and the highest correlation measure The corresponding single historical fault parameter matrix The basic fault parameter matrix is ​​determined as the cluster set of the historical fault types. .

7. The power grid fault analysis method based on artificial intelligence according to claim 6, characterized in that, The step of establishing a fault correlation prediction model corresponding to the fault type by performing parameter change correlation analysis relative to the baseline parameters based on the fault type baseline parameter information and combining the different single historical fault data information under the historical fault type cluster, includes: Based on the fault type baseline parameter information, determine the single historical fault parameter matrix corresponding to the single historical fault data information under the historical fault type cluster set. The remaining term historical fault parameter matrix is ​​calibrated. u represents the number of the different single historical fault data information remaining after the fault type baseline parameter information is removed from the historical fault type cluster; For different residual terms of historical fault parameter matrices Determine the relative basic fault parameter matrix Relative fault parameter change matrix ,in, ; Based on the fault type baseline parameter information, determine the single historical monitoring parameter matrix corresponding to other single historical fault data digitization information under the historical fault type cluster set. The remaining term is defined as the historical monitoring parameter matrix. ; For different historical monitoring parameter matrices of the remaining sub-items Determine the relative basic monitoring parameter matrix relative monitoring parameter change matrix ,in, ; The relative fault parameter change matrix corresponding to the same single historical fault data information. and the relative monitoring parameter change matrix The connections are used to form different information on the relative parameter changes of individual faults; By combining the relative parameter change information of different single faults, the maximum relative monitoring parameter values ​​of different monitoring parameters are determined. and the minimum value of relative monitoring parameters This forms the relative variation range of the corresponding fault monitoring parameters. ; By aggregating different relative parameter change information from a single fault, a correlation analysis of parameter changes is performed, and the relative change range of the fault monitoring parameters corresponding to different monitoring parameters is considered. Establish a fault correlation prediction model corresponding to the fault type.

8. The power grid fault analysis method based on artificial intelligence according to claim 7, characterized in that, The data sets contain different information on the relative parameter changes of each single fault. Correlation analysis of these parameter changes is performed, and the relative change ranges of the fault monitoring parameters corresponding to different monitoring parameters are considered. Establishing a fault correlation prediction model corresponding to the fault type includes: For different single-fault relative parameter change information, the relative fault parameter change matrix is ​​generated. and the relative monitoring parameter change matrix The matrix elements are uniformly processed to form the corresponding relative fault uniform parameter change matrix. and relative monitoring unified parameter change matrix The maximum value of i is the largest of the maximum values ​​of k and n. A unified correlation prediction formula is established for the relative parameter changes of different single faults. : ， Where v represents the number of different fault types. This represents the relative fault prediction parameter matrix. This represents the matrix representing the relative real-time monitoring parameter changes. This represents the transpose of the matrix representing the changes in the relative real-time monitoring parameters. This represents a matrix of power-law correlation parameters. This represents the single-power correlation parameter matrix. Represents a constant parameter matrix; For different single-fault relative parameter change information, the corresponding relative fault unified parameter change matrix will be generated. As the correlation prediction formula The relative fault prediction parameter matrix in the matrix will include the relative monitoring unified parameter change matrix. As the correlation prediction formula The relative real-time monitoring parameter change matrix is ​​used to solve for the power-law correlation parameter matrix, the single-law correlation parameter matrix, and the constant parameter matrix. The solved correlation prediction formula The relative variation range of the fault monitoring parameters and the corresponding different monitoring parameters The fault correlation prediction model corresponding to the fault type is determined.

9. The power grid fault analysis method based on artificial intelligence according to claim 8, characterized in that, The process of acquiring real-time monitoring data, combining it with the fault correlation prediction model to perform fault prediction analysis, and forming real-time fault prediction results includes: Based on the real-time monitoring data, combined with the basic monitoring parameter matrix corresponding to different fault types... The corresponding real-time monitoring parameter change matrix is ​​determined; The relative variation range of fault monitoring parameters according to different fault types It was determined that the changes in all monitoring parameters in the real-time monitoring parameter change matrix belonged to the relative change range of the fault monitoring parameters corresponding to the fault type. The fault type is identified and labeled as a real-time predicted fault type; Obtain the fault correlation prediction model corresponding to the real-time predicted fault type, and combine it with the real-time monitoring parameter change matrix to determine the real-time fault parameter change information; Based on the real-time fault parameter change information, combined with the basic fault parameter matrix corresponding to the real-time predicted fault type, The parameter values ​​of different fault parameters are determined and real-time fault prediction results are generated.

10. An artificial intelligence-based power grid fault analysis system, employing the artificial intelligence-based power grid fault analysis method according to any one of claims 1-9, characterized in that, include: The data acquisition unit is used to acquire historical fault data and real-time monitoring data; The clustering processing unit is used to perform data clustering analysis on the historical fault data acquired by the data acquisition unit according to the fault type, and form historical fault cluster data; The training prediction unit is used to perform correlation analysis on the changes in parameters of the same type of fault based on the historical fault clustering data formed by the clustering processing unit, and to obtain a fault correlation prediction model. The real-time analysis unit is used to perform fault prediction analysis based on the real-time monitoring data obtained by the data acquisition unit and the fault correlation prediction model formed by the training prediction unit, and to form a real-time fault prediction result.

Images