Transformer substation equipment anomaly detection method and system based on extreme value theory, and medium

Abstract

The invention relates to a substation equipment anomaly detection method and system based on the extreme value theory and a medium, and the method comprises the steps: obtaining the monitoring data of substation equipment, and determining an anomaly detection threshold value delta i; calculating an extreme index prediction capability value gamma i based on an abnormal result of the substation equipment and an abnormal detection threshold value delta i; calculating a conventional index prediction capability value alpha i; determining a comprehensive prediction capability value # imgabs0 # of the base algorithm based on an extreme index prediction capability value gamma i and a conventional index prediction capability value alpha i, performing standardization operation on the comprehensive prediction capability value # imgabs1 # of the base algorithm to obtain a combined weight # imgabs2 #, and weighting the combined weight # imgabs3 # to a detection result of a reference method to perform abnormal data monitoring. Therefore, the accuracy and reliability of anomaly detection are improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of substation abnormality detection, and in particular to a substation equipment abnormality detection method, system and medium based on extreme value theory. Background Art

[0002] Anomaly detection of smart substation equipment is a key technology that is of great significance in identifying and predicting equipment or system failures.

[0003] Traditional anomaly detection methods are usually passive, triggering alerts only when indicators exceed preset thresholds or deviate significantly from historical data. However, this approach may not accurately predict extreme data, thus reducing the accuracy of anomaly detection. Summary of the invention

[0004] This application is made to solve the above problems, and aims to provide a substation equipment anomaly detection method, system and medium based on extreme value theory, which improves the accuracy and robustness of substation equipment anomaly detection by using an integrated learning algorithm based on extreme value theory. In order to achieve the above purpose, this application adopts the following scheme:

[0005] The present application provides a substation equipment abnormality detection method based on extreme value theory, and the substation equipment abnormality detection method comprises the following steps:

[0006] Step 1: Obtain monitoring data of substation equipment and determine the anomaly detection threshold δ i ;

[0007] Step 2: Based on the abnormal results of the substation equipment and the abnormal detection threshold δ i , calculate the extreme index prediction ability value γ i ;

[0008] Step 3: Calculate the predictive ability value α of the conventional index i ;

[0009] Step 4: Predict the capability value γ based on the extreme indicator i and the conventional index prediction ability value α i Determine the comprehensive predictive ability value of the base algorithm

[0010] Step 5: Comprehensive prediction capability value of the base algorithm Perform standardization operation to obtain the combined weight

[0011] Step 6: Combine the weights The detection result of the benchmark method is weighted to monitor abnormal data, wherein judging whether the detection result of the substation equipment is abnormal includes conventional detection and extreme detection.

[0012] Furthermore, the step 1 comprises:

[0013] Step 1.1: Get the indicator deviation sequence Specifically:

[0014] in Represents the prediction result of the ith base algorithm for the tth indicator With the real indicator y t Deviation;

[0015] Step 1.2: Determine the Hill graph statistic H based on the indicator deviation sequence i,t , specifically:

[0016] in is the ascending statistic of the deviation sequence, and the set of ascending statistics is

[0017] Step 1.3: Slope-based change point detection method determines the optimal change point in the Hill plot statistic The slope-based change point detection method is specifically as follows:

[0018] In the formula, F i (t) represents the slope-based change point detection function, a1 and a2 represent the intercept constants of the ascending statistics before and after the mutation point, respectively. and They are respectively expressed as the slopes before and after the mutation point, is the change point to be found, ε t is a random term, ε t ~N(0,1);

[0019] The optimal change point Specifically:

[0020]

[0021] Step 1.4: Convert the optimal change point to obtain the anomaly detection threshold δ i , specifically:

[0022]

[0023] Furthermore, the step 2 comprises:

[0024] Step 2.1: Construct the conditional distribution formula of the extreme value model Specifically:

[0025]

[0026] Step 2.2: Obtain the extreme value distribution function G based on the conditional distribution formula i (x), specifically:

[0027] Where x≥δ i , indicating that the indicator is greater than δ i data;

[0028] Step 2.3: performing parameter estimation on the shape parameter and scale parameter of the object index quantity in the extreme value distribution function;

[0029] Step 2.4: Based on the above parameter estimation results, the maximum possible extreme deviation is obtained, which is:

[0030]

[0031] In the formula, β i and σ i They are the shape parameter and scale parameter of the object indicator quantity;

[0032] Step 2.5: Based on the above extreme value deviation, we can get the prediction ability of extreme indicators γ i , specifically:

[0033] Where μ represents the regulatory factor of predictive ability, μ>0, p i Indicates the proportion of Var of the i-th base algorithm in all base algorithms.

[0034] Furthermore, the step 2.3 includes:

[0035] Step 2.31: Transform the extreme value distribution function to obtain the intermediate calculation formula, which is:

[0036]

[0037] Step 2.32: Substituting the empirical estimate into the above intermediate calculation formula, we can get the following formula:

[0038]

[0039] In the formula Indicates the number of indicators exceeding the threshold;

[0040] Step 2.33: Further calculation yields

[0041]

[0042] Step 2.34: Use the probability weight moment estimation to estimate the parameters of the above formula, specifically:

[0043] In the formula

[0044] Furthermore, the p i Specifically:

[0045] In the formula

[0046] Furthermore, the step 3 comprises:

[0047] Step 3.1: Get the deviations that do not exceed the threshold to form a set Specifically:

[0048] In the formula Indicates the number of indicators that do not exceed the threshold;

[0049] Step 3.2: Evaluate the ability of the base algorithm to predict conventional indicators by using the mean square error function MSE i , the mean square error function MSE is specifically:

[0050]

[0051] The ability of predicting conventional indicators obtained is α i Specifically:

[0052] In the formula

[0053] Furthermore, the step 4 comprises:

[0054] The calculation formula of the comprehensive prediction ability value is:

[0055] Where τ t Indicates the ranking ratio of the t-th data in the data that have appeared.

[0056] Furthermore, the step 5 comprises:

[0057] The calculation formula of the combined weight is:

[0058]

[0059] The present application also provides a substation equipment abnormality detection system based on extreme value theory, and the integrated learning system includes:

[0060] A threshold determination module is configured to determine an anomaly detection threshold δ i ;

[0061] The extreme index prediction capability value calculation module is configured to calculate the extreme index prediction capability value γ i ;

[0062] A conventional indicator prediction capability value calculation module is configured to calculate the conventional indicator prediction capability value α i ;

[0063] A comprehensive prediction capability value determination module is configured to determine the comprehensive prediction capability value of the base algorithm based on the extreme indicator prediction capability value and the conventional indicator prediction capability value.

[0064] The weight calculation module is configured to perform a standardization operation on the comprehensive prediction ability value to obtain a combined weight

[0065] The anomaly detection module is configured to add the above-mentioned combined weight to the detection result of the benchmark method to detect abnormal data.

[0066] The present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, a substation equipment abnormality detection method as described in any one of the above items is implemented.

[0067] Compared with the prior art, the present invention has the following beneficial effects:

[0068] This application determines the comprehensive prediction ability value of the base algorithm by calculating the extreme indicator prediction ability value and the conventional indicator prediction ability value, and performs a standard operation on the comprehensive prediction ability value to obtain a combined weight. The combined weight is weighted to the detection result of the benchmark method to monitor abnormal data, which can consider both conventional detection and extreme detection, thereby improving the accuracy of abnormal detection, not only effectively identifying and predicting equipment or system failures, but also adapting to changes in extreme data and improving the reliability of abnormal detection.

[0069] This application has made significant breakthroughs in the field of abnormality detection of substation equipment, improved the accuracy of abnormality analysis and prediction, is suitable for processing complex data relationships, improves the operating efficiency of substation equipment fault analysis, effectively reduces the risk of overfitting, and has significant practical value and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0071] Figure 1 It is a system block diagram of a substation equipment abnormality detection method based on extreme value theory involved in an embodiment of the present application;

[0072] Figure 2 It is a principle block diagram of abnormality detection of substation equipment based on extreme value theory involved in an embodiment of the present application;

[0073] Figure 3 It is a flow chart of a substation equipment abnormality detection method based on extreme value theory involved in an embodiment of the present application;

[0074] Figure 4 is a prediction deviation diagram of an embodiment of the present application;

[0075] Figure 5 is the prediction deviation plot of the existing benchmark methods; DETAILED DESCRIPTION

[0076] The specific implementation methods of the present application are described in detail below with reference to the accompanying drawings.

[0077] It should be noted that, in the absence of conflict, the embodiments and technical features in the embodiments of the present application can be combined with each other, and the detailed description in the specific implementation method should be understood as an explanation of the purpose of the present application and should not be regarded as an improper limitation on the present application.

[0078] It should be understood that the orientation or position relationship is based on the orientation or position relationship shown in the drawings. These orientation terms are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present application.

[0079] Fault analysis technology for smart substation equipment is a key technology that is of great significance in identifying and predicting equipment or system failures. The effective implementation of this technology requires multiple steps, from abnormality detection to fault diagnosis and then to fault prediction. The following is a brief introduction to these steps:

[0080] Anomaly detection: First, the various indicators of the equipment or system need to be monitored and detected to determine whether they are abnormal. These indicators may cover parameters such as temperature, pressure, current, voltage, etc. Anomalies can be indicators exceeding the preset threshold or showing significant deviations compared with historical data.

[0081] Feature extraction: Once an anomaly is found, the next step is to extract features from these abnormal indicators. These features may include statistics such as the mean, variance, and rate of change of the indicator, or more complex features such as spectral features, time-frequency features, etc.

[0082] Model training: The extracted features will be used as input to train the machine learning model. Here, one of the commonly used methods is the support vector machine (SVM), which is a supervised learning algorithm used for classification and regression analysis. The SVM model can detect anomalies in new data by learning the characteristic patterns under normal operating conditions.

[0083] Fault diagnosis: The trained model can be used to classify new data to determine whether the device or system is in a normal state or has a fault. If the model identifies certain features that match known failure modes, further fault diagnosis and treatment can be performed.

[0084] Through the above steps, fault analysis technology can help detect potential faults of smart substation equipment in advance and take corresponding maintenance or repair measures to ensure its normal operation and reliability.

[0085] In view of this, an embodiment of the present application proposes a substation equipment abnormality detection method based on extreme value theory, which is based on ensemble learning of extreme value theory. Figure 1 is a system block diagram of a substation equipment abnormality detection method based on extreme value theory involved in an embodiment of the present application, Figure 2 is a principle block diagram of abnormality detection of substation equipment based on extreme value theory involved in an embodiment of the present application, Figure 3 is a flow chart of a method for detecting abnormality of substation equipment based on extreme value theory according to an embodiment of the present application, such as Figures 1 to 3 As shown, the substation equipment abnormality detection method includes the following steps:

[0086] Step 1: Obtain monitoring data of substation equipment and determine the anomaly detection threshold δ i ;

[0087] Step 2: Based on the abnormal results of the substation equipment and the abnormal detection threshold δ i , calculate the extreme index prediction ability value γ i ;

[0088] Step 3: Calculate the predictive ability value α of the conventional index i ;

[0089] Step 4: Predict the capability value γ based on the extreme indicator i and the conventional index prediction ability value α iDetermine the comprehensive predictive ability value of the base algorithm

[0090] Step 5: Comprehensive prediction capability value of the base algorithm Perform standardization operation to obtain the combined weight

[0091] Step 6: Combine the weights The detection result of the benchmark method is weighted to monitor abnormal data, wherein judging whether the detection result of the substation equipment is abnormal includes conventional detection and extreme detection.

[0092] Specifically, first collect the real-time monitoring data of each device in the substation, which may include parameters such as current, voltage, temperature, vibration, insulation resistance, etc. Use statistical analysis methods, such as historical data analysis and time series analysis, to determine the appropriate anomaly detection threshold. These thresholds should be set based on the normal working range of the equipment and historical anomaly data. For each monitored data point, compare its relationship with the threshold to determine whether it exceeds the preset anomaly threshold, and apply statistical models in extreme value theory, such as generalized extreme value distribution (Gumbel distribution), Freescheit distribution, etc., to predict the probability of extreme events and calculate the prediction ability value of extreme indicators. Calculate the prediction ability value of conventional indicators. In addition to extreme indicators, use machine learning algorithms such as support vector machine (SVM), random forest (RF) or artificial neural network (ANN) to model conventional indicators and calculate their prediction ability values. Combine the prediction ability values ​​of extreme indicators and conventional indicators, and use weighted average or other fusion techniques to obtain a comprehensive prediction ability value. The calculation of the comprehensive prediction ability value can reflect the importance of different indicators for anomaly detection, thereby adjusting the sensitivity of the algorithm to different types of anomalies.

[0093] The comprehensive predictive ability values ​​are standardized to obtain the combined weights. The comprehensive predictive ability values ​​are standardized to ensure that the weights are distributed between 0 and 1. The standardized weights reflect the contribution of each algorithm in the comprehensive model, and these weights are applied to the detection results of the benchmark method.

[0094] The combined weights are added to the detection results of the benchmark method to monitor abnormal data, and the results of the benchmark detection method are adjusted using the combined weights obtained above to improve the monitoring sensitivity of abnormal data. The monitoring results need to comprehensively consider conventional detection and extreme detection in order to more comprehensively judge the abnormal status of substation equipment.

[0095] It should be noted that in the overall framework of fault analysis, the embodiments of the present application have been optimized and improved for intelligent indicator prediction and classification models. On the one hand, traditional anomaly detection methods are usually passive, and alarms can only be triggered when the indicator exceeds a preset threshold or deviates significantly from historical data. However, this method may not accurately predict extreme data, thereby reducing the accuracy of anomaly detection. The improvement proposed in this application can more actively predict the abnormal values ​​of indicators by adopting an ensemble learning algorithm based on extreme value theory. The ensemble learning algorithm based on extreme value theory combines an in-depth understanding of data distribution and the idea of ​​ensemble learning. It combines the prediction results of multiple models and uses the anomaly detection method in extreme value theory to improve the ability to accurately predict abnormal values. This method can effectively capture extreme situations in data distribution and detect anomalies at an early stage, thereby improving the accuracy and efficiency of fault diagnosis. Through the above improvements, the present application can significantly improve the efficiency and accuracy of fault diagnosis, and provide important technical support for the stable operation and safety of key systems such as smart substations.

[0096] In some embodiments, step 1 comprises:

[0097] Step 1.1: Get the indicator deviation sequence Specifically:

[0098]

[0099] in, Represents the prediction result of the ith base algorithm for the tth indicator With the real indicator y t Deviation;

[0100] Step 1.2: Determine the Hill graph statistic H based on the indicator deviation sequence i,t , specifically:

[0101]

[0102] in is the ascending statistic of the deviation sequence, and the set of ascending statistics is

[0103] Step 1.3: Slope-based change point detection method determines the optimal change point in the Hill plot statistic The slope-based change point detection method is specifically as follows:

[0104]

[0105] In the formula, F i(t) represents the slope-based change point detection function, a1 and a2 represent the intercept constants of the ascending statistics before and after the mutation point, respectively. and They are respectively expressed as the slopes before and after the mutation point, is the change point to be found, ε t is a random term, ε t ~N(0,1);

[0106] The optimal change point Specifically:

[0107]

[0108] Step 1.4: Convert the optimal change point to obtain the anomaly detection threshold δ i , specifically:

[0109]

[0110] Specifically, when considering whether an indicator is an extreme indicator, what is considered is not the attribute of the indicator itself, but the deviation between the indicator predicted by the base algorithm and the true value of the indicator. Therefore, the sequence of deviations of the base algorithm used to judge whether an indicator is an extreme indicator can be carried out using the relationship (1.1).

[0111] The threshold is determined based on the deviation sequence. The indicators in the deviation sequence that exceed the threshold are defined as extreme indicators, and those that are below the threshold are defined as normal indicators. The Hill graph determines the threshold by finding the maximum value of the Hill statistic in a stable linear state. The Hill statistic refers to the relationship (1.2), and the optimal threshold is further obtained through the relationships (1.3), (1.4), (1.5), and (1.6).

[0112] In some embodiments, step 2 comprises:

[0113] Step 2.1: Construct the conditional distribution formula of the extreme value model Specifically:

[0114]

[0115] Step 2.2: Obtain the extreme value distribution function G based on the conditional distribution formula i (x), specifically:

[0116]

[0117] Where x≥δ i , indicating that the indicator is greater than δ i data;

[0118] Step 2.3: performing parameter estimation on the shape parameter and scale parameter of the object index quantity in the extreme value distribution function;

[0119] Step 2.4: Based on the above parameter estimation results, the maximum possible extreme deviation is obtained, which is:

[0120]

[0121] In the formula, β i and σ i They are the shape parameter and scale parameter of the object indicator quantity;

[0122] Step 2.5: Based on the above extreme value deviation, we can get the prediction ability of extreme indicators γ i , specifically:

[0123]

[0124] Where μ represents the regulatory factor of predictive ability, μ>0, p i Indicates the proportion of Var of the i-th base algorithm in all base algorithms.

[0125] The step 2.3 comprises:

[0126] Step 2.31: Transform the extreme value distribution function to obtain the intermediate calculation formula, which is:

[0127]

[0128] Step 2.32: Substituting the empirical estimate into the above intermediate calculation formula, we can get the following formula:

[0129]

[0130] In the formula Indicates the number of indicators exceeding the threshold;

[0131] Step 2.33: Further calculation yields

[0132]

[0133] Step 2.34: Use the probability weight moment estimation to estimate the parameters of the above formula, specifically:

[0134]

[0135] In the formula

[0136] The ability is evaluated based on the maximum possible extreme deviation of the base algorithm at a given confidence level. According to the relationship (1.8), the maximum possible extreme deviation is determined, and then the prediction ability of the extreme index γ can be obtained. i .

[0137] In some embodiments, the p i Specifically:

[0138]

[0139] In the formula The proportion of Var of the i-th base algorithm in all base algorithms is determined by (1.16).

[0140] In some embodiments, step 3 comprises:

[0141] Step 3.1: Get the deviations that do not exceed the threshold to form a set Specifically:

[0142]

[0143] In the formula Indicates the number of indicators that do not exceed the threshold;

[0144] Step 3.2: Evaluate the ability of the base algorithm to predict conventional indicators by using the mean square error function MSE i , the mean square error function MSE is specifically:

[0145]

[0146] The ability of predicting conventional indicators obtained is α i Specifically:

[0147]

[0148] Based on the relations (1.18) and (1.19), the ability α to predict conventional indicators is obtained i ,in

[0149] The step 4 comprises:

[0150] The calculation formula of the comprehensive prediction ability value is:

[0151] Where τ t Indicates the ranking ratio of the t-th data in the data that have appeared.

[0152] In some embodiments, step 5 comprises:

[0153] The calculation formula of the combined weight is:

[0154]

[0155] This application uses mean square error to evaluate the ability of the base algorithm to predict conventional indicators, and the extreme case of the indicator τ t To reflect the LPEE algorithm's preference for predictive ability, where τ t It represents the ranking ratio of the t-th data in the data that have appeared, and obtains the comprehensive prediction ability of the base algorithm at time t.

[0156] In order to better understand the embodiments of the present application, the prediction accuracy of anomaly detection using an ensemble learning algorithm based on extreme value theory is first evaluated in conjunction with the present invention.

[0157] The experimental results of the present embodiment are as follows: Figure 4 The distribution of the prediction deviations was evaluated in detail, where the baseline method showed a significant fat-tailed distribution, especially around the deviation value of about 7. This distribution characteristic shows that the baseline method has difficulty in accurately predicting extreme values ​​in some cases, resulting in uneven and unstable deviation distribution. In contrast, the method used in the embodiment of the present application showed significant improvement. Figure 5 As shown, in the prediction results, the deviations are mainly concentrated between 0 and 2.5, and the distribution is more concentrated and symmetrical. This means that the embodiment of the present application can predict data more accurately and has better coping capabilities for extreme situations that deviate from the normal range. By limiting the deviations to a smaller range, the embodiment of the present application effectively eliminates the original fat-tail distribution phenomenon, making the prediction results more stable and reliable.

[0158] This improvement is not only reflected in the graph, but also verified in the actual numerical indicators. Compared with the benchmark method, the prediction results of the present invention are obviously closer to the actual values, and the mean square error is significantly reduced. This further verifies the advantages of the present invention in terms of prediction accuracy and stability, and provides reliable technical support for fault diagnosis and prediction.

[0159] The embodiment of the present application provides a substation equipment abnormality detection system based on extreme value theory, and the integrated learning system includes:

[0160] A threshold determination module is configured to determine an anomaly detection threshold δ i ;

[0161] The extreme index prediction capability value calculation module is configured to calculate the extreme index prediction capability value γ i ;

[0162] A conventional indicator prediction capability value calculation module is configured to calculate the conventional indicator prediction capability value αi ;

[0163] A comprehensive prediction capability value determination module is configured to determine the comprehensive prediction capability value of the base algorithm based on the extreme indicator prediction capability value and the conventional indicator prediction capability value.

[0164] The weight calculation module is configured to perform a standardization operation on the comprehensive prediction ability value to obtain a combined weight

[0165] The anomaly detection module is configured to add the above-mentioned combined weight to the detection result of the benchmark method to detect abnormal data.

[0166] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, a method for detecting abnormality of substation equipment as described in any one of the above items is implemented.

[0167] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit the same. Although the present application has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to replace some of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions claimed to be protected by the present application.

Claims

1. A method for detecting abnormality of substation equipment based on extreme value theory, characterized in that: The substation equipment abnormality detection method comprises the following steps: Step 1: Obtain monitoring data of substation equipment and determine the anomaly detection threshold δ i ; Step 2: Based on the abnormal results of the substation equipment and the abnormal detection threshold δ i , calculate the extreme index prediction ability value γ i ; Step 3: Calculate the predictive ability value α of the conventional index i ; Step 4: Predict the capability value γ based on the extreme indicator i and the conventional index prediction ability value α i Determine the comprehensive predictive ability value of the base algorithm Step 5: Comprehensive prediction capability value of the base algorithm Perform standardization operation to obtain the combined weight Step 6: Combine the weights The detection result of the benchmark method is weighted to monitor abnormal data, wherein judging whether the detection result of the substation equipment is abnormal includes conventional detection and extreme detection.

2. The method for detecting abnormality of substation equipment according to claim 1, characterized in that: The step 1 comprises: Step 1.1: Get the indicator deviation sequence Specifically: in Represents the prediction result of the ith base algorithm for the tth indicator With the real indicator y t Deviation; Step 1.2: Determine the Hill graph statistic H based on the indicator deviation sequence i,t , specifically: in is the ascending statistic of the deviation sequence, and the set of ascending statistics is Step 1.3: Slope-based change point detection method determines the optimal change point in the Hill plot statistic The slope-based change point detection method is specifically as follows: In the formula, F i (t) represents the slope-based change point detection function, a1 and a2 represent the intercept constants of the ascending statistics before and after the mutation point, respectively. and They are respectively expressed as the slopes before and after the mutation point, is the change point to be found, ε t is a random term, ε t ~N(0,1); The optimal change point Specifically: Step 1.4: Convert the optimal change point to obtain the anomaly detection threshold δ i , specifically:

3. The method for detecting abnormality of substation equipment according to claim 2, characterized in that: The step 2 comprises: Step 2.1: Construct the conditional distribution formula of the extreme value model Specifically: Step 2.2: Obtain the extreme value distribution function G based on the conditional distribution formula i (x), specifically: Where x≥δ i , indicating that the indicator is greater than δ i data; Step 2.3: performing parameter estimation on the shape parameter and scale parameter of the object index quantity in the extreme value distribution function; Step 2.4: Based on the above parameter estimation results, the maximum possible extreme deviation is obtained, which is: In the formula, β i and σ i They are the shape parameter and scale parameter of the object indicator quantity; Step 2.5: Based on the above extreme value deviation, we can get the prediction ability of extreme indicators γ i , specifically: Where μ represents the regulatory factor of predictive ability, μ>0, p i Indicates the proportion of Var of the i-th base algorithm in all base algorithms.

4. The method for detecting abnormality of substation equipment according to claim 3, characterized in that: The step 2.3 includes: Step 2.31: Transform the extreme value distribution function to obtain the intermediate calculation formula, which is: Step 2.32: Substituting the empirical estimate into the above intermediate calculation formula, we can get the following formula: In the formula Indicates the number of indicators exceeding the threshold; Step 2.33: Further calculation yields Step 2.34: Use the probability weight moment estimation to estimate the parameters of the above formula, specifically: In the formula 5. The method for detecting abnormality of substation equipment according to claim 3, characterized in that: The p i Specifically: In the formula 6. The method for detecting abnormality of substation equipment according to claim 3, characterized in that: The step 3 comprises: Step 3.1: Get the deviations that do not exceed the threshold to form a set Specifically: In the formula Indicates the number of indicators that do not exceed the threshold; Step 3.2: Evaluate the ability of the base algorithm to predict conventional indicators by using the mean square error function MSE i , the mean square error function MSE is specifically: The ability of predicting conventional indicators obtained is α i Specifically: In the formula 7. The method for detecting abnormality of substation equipment according to claim 1, characterized in that: The step 4 comprises: The calculation formula of the comprehensive prediction ability value is: Where τ t Indicates the ranking ratio of the t-th data in the data that have appeared.

8. The method for detecting abnormality of substation equipment according to claim 1, characterized in that: The step 5 comprises: The calculation formula of the combined weight is:

9. A substation equipment abnormality detection system based on extreme value theory, characterized in that: The integrated learning system comprises: A threshold determination module is configured to determine an anomaly detection threshold δ i ; The extreme index prediction capability value calculation module is configured to calculate the extreme index prediction capability value γ i ; A conventional indicator prediction capability value calculation module is configured to calculate the conventional indicator prediction capability value α i ; A comprehensive prediction capability value determination module is configured to determine the comprehensive prediction capability value of the base algorithm based on the extreme indicator prediction capability value and the conventional indicator prediction capability value. The weight calculation module is configured to perform a standardization operation on the comprehensive prediction ability value to obtain a combined weight The anomaly detection module is configured to add the above-mentioned combined weight to the detection result of the benchmark method to detect abnormal data.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for detecting abnormality of substation equipment according to any one of claims 1 to 8 is implemented.

Images