Method for graded early warning of risk in distribution transformer area, system, device, storage medium, and program product

Abstract

A method for graded early warning of a risk in a distribution transformer area, a system, a device, a storage medium, and a program product. The method comprises: performing feature extraction on target monitoring data of a distribution transformer area collected on the basis of a first data collection time interval, so as to obtain a first transformer area data feature; on the basis of the first transformer area data feature, performing correlation analysis on a preset feature index to obtain a first key early warning index; using the first key early warning index to perform feature index fusion to obtain first new feature data; on the basis of the first new feature data, using an early warning model to predict to obtain an early warning level for the distribution transformer area; on the basis of the early warning level for the distribution transformer area, adjusting the first data collection time interval to obtain a second data collection time interval; and re-collecting target monitoring data on the basis of the second data collection time interval, and re-determining the early warning level for the distribution transformer area on the basis of the re-collected target monitoring data.

Description

Methods, systems, equipment, storage media, and program products for risk classification and early warning of power distribution areas.

[0001] Cross-reference to related applications

[0002] This application is based on and claims priority to Chinese Patent Application No. 202411587987.1, filed on November 8, 2024, entitled “A Method and System for Risk Classification and Early Warning of Distribution Areas”, the entire contents of which are incorporated herein by reference. Technical Field

[0003] This application belongs to the field of power distribution network operation monitoring technology, specifically relating to a method, system, equipment, storage medium and program product for risk classification and early warning of power distribution substations. Background Technology

[0004] With the significant increase in the number of electric vehicles, large-capacity electric vehicle charging stations are gradually being connected to the power distribution network, providing convenient energy services for charging and discharging electric vehicles. However, due to the lag in the monitoring equipment and methods used in some power distribution network areas, the current monitoring capabilities for distribution areas with electric vehicle charging stations are insufficient, making it difficult to meet the demand for efficient monitoring. Furthermore, the concentrated charging of a large number of electric vehicles during operation may lead to overload, voltage deviation, and other deterioration issues in the corresponding distribution areas. At the same time, the power distribution network lacks efficient methods for monitoring and early warning of distribution areas, making it difficult to meet the monitoring and early warning requirements for ensuring the high-quality operation of the power distribution network.

[0005] Current research addresses the issue of monitoring the operational status of distribution network substations. The conventional method involves collecting relevant electrical quantities at 15-minute intervals and analyzing the operational status to generate alarms. While this method is suitable for monitoring and managing substations with relatively low operational requirements, it struggles to provide timely warnings of operational deterioration in substations containing electric vehicle charging stations. Without more efficient monitoring and accurate early warning systems for substations with electric vehicle charging stations, the operational status of these substations will be severely affected, restricting the energy supply level and reliability of the substations where charging stations are located. Summary of the Invention

[0006] To overcome the problems existing in the above-mentioned related technologies, this application provides a method, system, device, storage medium and program product for risk classification and early warning of distribution radio areas.

[0007] According to a first aspect of the embodiments of this application, a method for risk classification and early warning of distribution radio areas is provided, including:

[0008] Feature extraction is performed on the target monitoring data of the distribution station area to obtain the first data feature of the distribution station area; the target monitoring data of the distribution station area is collected based on the first data acquisition time interval;

[0009] Based on the data characteristics of the first transformer area, a correlation analysis is performed on the preset feature indicators to obtain the first key early warning indicator.

[0010] The first key early warning indicator is used to fuse feature indicators to obtain the first new feature data;

[0011] Based on the first new feature data, the warning level of the distribution radio area is predicted using the warning model.

[0012] Based on the warning level of the distribution radio area, the first data acquisition time interval is adjusted to obtain the second data acquisition time interval;

[0013] The target monitoring data is re-acquired based on the second data acquisition time interval, and the warning level of the distribution radio station is re-determined based on the re-acquired target monitoring data.

[0014] In some embodiments, the step of extracting features from the target monitoring data of the distribution substation collected based on the first data acquisition time interval to obtain the first substation data features includes: cleaning the target monitoring data; and extracting data corresponding to preset feature indicators from the cleaned target monitoring data to obtain the first substation data features.

[0015] In some embodiments, the step of performing correlation analysis on preset feature indicators based on the data characteristics of the first transformer area to obtain the first key early warning indicator includes: performing correlation analysis on the preset feature indicators using a gray-scale correlation analysis method based on the data characteristics of the first transformer area to obtain the first key early warning indicator.

[0016] In some embodiments, the step of performing correlation analysis on the preset feature indicators based on the data characteristics of the first transformer area to obtain the first key early warning indicator includes: constructing a first comparison series using the data characteristics of the first transformer area; constructing a first reference series using the collected first historical monitoring data; standardizing both the first comparison series and the first reference series; for each preset feature indicator, determining the absolute difference between the standardized first comparison series corresponding to the preset feature indicator and the standardized first reference series corresponding to the feature indicator; for each preset feature indicator, determining the gray-scale correlation degree between the standardized first comparison series corresponding to the preset feature indicator and the standardized first reference series corresponding to the feature indicator using the absolute difference; and selecting the preset feature indicator corresponding to the gray-scale correlation degree that meets the preset conditions as the first key early warning indicator.

[0017] In some embodiments, constructing a first reference sequence using the collected first historical monitoring data includes: cleaning the first historical monitoring data; standardizing the cleaned first historical monitoring data corresponding to the preset feature indicators to obtain standardized first historical monitoring data corresponding to the preset feature indicators; setting the minimum value among the standardized first historical monitoring data corresponding to each preset feature indicator as the optimal value and the maximum value among the standardized first historical monitoring data corresponding to each preset feature indicator as the worst value; and constructing the first reference sequence using the optimal and worst values ​​corresponding to the preset feature indicators.

[0018] In some embodiments, determining the grayscale correlation between the standardized first comparison series corresponding to the feature index and the standardized first reference series corresponding to the feature index using the absolute difference value of the feature index includes: determining the maximum absolute difference and the minimum absolute difference between the standardized first comparison series corresponding to the preset feature index and the standardized first reference series corresponding to the preset feature index based on the absolute difference value; and determining the grayscale correlation between the standardized first comparison series and the standardized first reference series corresponding to each preset feature index using the maximum absolute difference and the minimum absolute difference value.

[0019] In some embodiments, the step of fusing feature indicators using the first key early warning indicator to obtain the first new feature data includes: extracting data corresponding to the first key early warning indicator from the cleaned target monitoring data to construct a first feature set matrix; and constructing the first new feature data based on the first feature set matrix using principal component analysis.

[0020] In some embodiments, constructing the first new feature data based on the first feature set matrix using principal component analysis includes: calculating a first covariance matrix using the first feature set matrix; calculating the cumulative contribution rate of the first p features in the first covariance matrix; and constructing the first new feature data using the first p features when the cumulative contribution rate of the first p features is greater than or equal to the expected value of the cumulative contribution rate. Here, p is a positive integer, p≤k, and k is the total number of features in the first covariance matrix.

[0021] In some embodiments, adjusting the first data acquisition time interval according to the warning level of the distribution transformer area to obtain a second data acquisition time interval includes: when the warning level is the fourth warning level or the number of electric vehicles entering the distribution transformer area reaches a first percentage of the charging station capacity, the second data acquisition time interval is a first multiple of the first data acquisition time interval; when the warning level is the third warning level or the number of electric vehicles entering the distribution transformer area reaches a second percentage of the charging station capacity, the second data acquisition time interval is a second multiple of the first data acquisition time interval; when the warning level is lower than or equal to the second warning level and the number of electric vehicles entering the distribution transformer area is less than a second percentage of the charging station capacity, the second data acquisition time interval is a third multiple of the first data acquisition time interval; when the warning level is lower than or equal to the first warning level and the number of electric vehicles entering the distribution transformer area is less than a second percentage of the charging station capacity, the second data acquisition time interval is the same as the first data acquisition time interval.

[0022] In some embodiments, the process of establishing the early warning model includes: collecting second historical monitoring data of a distribution substation; extracting features from the second historical monitoring data to obtain second substation data features; performing correlation analysis on preset feature indicators based on the second substation data features to obtain second key early warning indicators; fusing feature indicators using the second key early warning indicators to obtain second new feature data; determining the historical early warning level of the distribution substation corresponding to the second new feature data; constructing a dataset using the second new feature data and the historical early warning level of the distribution substation, and training a long and short neural network model using the dataset to obtain the early warning model.

[0023] In some embodiments, the step of extracting features from the second historical monitoring data to obtain second transformer area data features includes: cleaning the second historical monitoring data; and extracting data corresponding to preset feature indicators from the cleaned second historical monitoring data to obtain second transformer area data features.

[0024] In some embodiments, the step of performing correlation analysis on preset feature indicators based on the data characteristics of the second transformer area to obtain the second key early warning indicator includes: performing correlation analysis on preset feature indicators using a gray-scale correlation analysis method based on the data characteristics of the second transformer area to obtain the second key early warning indicator.

[0025] In some embodiments, the step of performing correlation analysis on the preset feature indicators based on the data characteristics of the second transformer area to obtain the second key early warning indicator includes: constructing a second comparison series using the data characteristics of the second transformer area; constructing a second reference series using the collected third historical monitoring data; standardizing both the second comparison series and the second reference series; for each preset feature indicator, determining the absolute difference between the standardized second comparison series corresponding to the preset feature indicator and the standardized second reference series corresponding to the feature indicator; for each preset feature indicator, determining the gray-scale correlation degree between the standardized second comparison series corresponding to the preset feature indicator and the standardized second reference series corresponding to the feature indicator using the absolute difference; selecting the feature indicator corresponding to the gray-scale correlation degree that meets preset conditions as the second key early warning indicator; wherein the collection time of the third historical monitoring data is earlier than the collection time of the second historical monitoring data.

[0026] In some embodiments, constructing a second reference sequence using the collected third historical monitoring data includes: cleaning the third historical monitoring data; standardizing the cleaned third historical monitoring data corresponding to the preset feature indicators to obtain standardized third historical monitoring data corresponding to the preset feature indicators; setting the minimum value among the standardized third historical monitoring data corresponding to each preset feature indicator as the optimal value and the maximum value among the standardized third historical monitoring data corresponding to each preset feature indicator as the worst value; and constructing the second reference sequence using the optimal and worst values ​​corresponding to the preset feature indicators.

[0027] In some embodiments, determining the grayscale correlation between the standardized second comparison series corresponding to the preset feature index and the standardized second reference series corresponding to the feature index using the absolute difference corresponding to the feature index includes: determining the maximum absolute difference and the minimum absolute difference between the standardized second comparison series corresponding to the preset feature index and the standardized second reference series corresponding to the preset feature index based on the absolute difference; and determining the grayscale correlation between the standardized second comparison series and the standardized second reference series corresponding to each preset feature index using the maximum absolute difference and the minimum absolute difference.

[0028] In some embodiments, the step of fusing feature indicators using the second key early warning indicator to obtain second new feature data includes: extracting data corresponding to the second key early warning indicator from the second historical monitoring data after data cleaning, and constructing a second feature set matrix; and constructing the second new feature data based on the second feature set matrix using principal component analysis.

[0029] In some embodiments, constructing the second new feature data based on the second feature set matrix using principal component analysis includes: determining a second covariance matrix using the second feature set matrix; determining the cumulative contribution rate of the first p' features in the second covariance matrix; and constructing the second new feature data using the first p' features in the second covariance matrix when the cumulative contribution rate of the first p' features is greater than or equal to the expected value of the cumulative contribution rate; wherein p' is a positive integer, p'≤k', and k' is the total number of features in the second covariance matrix.

[0030] In some embodiments, determining the historical warning level of the distribution substation corresponding to the second new feature data includes: extracting valid data from the second new feature data based on preset standard limits for normal operation and preset standard limits for fault operation of the substation corresponding to the second key warning indicator, to obtain processed second new feature data; performing feature clustering on the processed second new feature data using the K-means clustering method based on the second key warning indicator, to obtain clustered second new feature data; analyzing the extreme values ​​of change corresponding to each second key warning indicator using the interval prediction method based on the clustered second new feature data; predicting the sample variance of the Bootstrap sample corresponding to each second key warning indicator using the Bootstrap algorithm based on the clustered second new feature data and the extreme values ​​of change corresponding to each second key warning indicator; calculating the upper and lower limits of each second key warning indicator using the sample variances corresponding to each second key warning indicator; and determining the historical warning level of the distribution substation corresponding to the second new feature data based on the upper and lower limits of each second key warning indicator.

[0031] In some embodiments, determining the historical warning level of the distribution station corresponding to the second new feature data based on the upper and lower limits of each of the second key warning indicators includes: when the upper and lower limits of the second key warning indicators fall within the fluctuation range corresponding to a first confidence threshold, the historical warning level of the distribution station is a first warning level; when the upper and lower limits of the second key warning indicators fall within the fluctuation range corresponding to a second confidence threshold, the historical warning level of the distribution station is a second warning level; when the upper and lower limits of the second key warning indicators fall within the fluctuation range corresponding to a third confidence threshold, the historical warning level of the distribution station is a third warning level; and when the upper and lower limits of the second key warning indicators fall within the fluctuation range corresponding to a fourth confidence threshold, the historical warning level of the distribution station is a fourth warning level.

[0032] In some embodiments, training the long-short neural network model using the dataset to obtain the early warning model includes: dividing the dataset into a training set, a validation set, and a test set; training the long-short neural network model using the training set to obtain a trained long-short neural network model; validating the trained long-short neural network model using the validation set; if the validation is successful, evaluating the performance of the trained long-short neural network model using the test set; if the validation fails, adjusting the hyperparameters of the long-short neural network model and monitoring overfitting, and retraining the long-short neural network model until successful validation; evaluating the performance of the trained long-short neural network model using the test set; if the performance evaluation of the trained long-short neural network model is successful, the trained long-short neural network model is the early warning model; if the performance evaluation of the trained long-short neural network model fails, the trained long-short neural network model does not contain the characteristics of the transformer area.

[0033] In some embodiments, the formula for calculating the first covariance matrix includes: B = X T X

[0034] The formula for calculating the cumulative contribution rate of the first p features in the first covariance matrix includes:

[0035] In the above formula, X is the first feature set matrix, T is the transpose, and B is the first covariance matrix; i∈[1,k], k is the total number of features in the first covariance matrix, p≤k; α P λ represents the cumulative contribution rate of the first p features in the first covariance matrix. i Let be the eigenvalue of the i-th feature in the first covariance matrix.

[0036] In some embodiments, the formula for calculating the second covariance matrix includes: B′=X′ T X′

[0037] The formula for calculating the cumulative contribution rate of the first p' features in the second covariance matrix includes:

[0038] In the above formula, X' is the second feature set matrix, T is the transpose, and B' is the second covariance matrix; i′∈[1,k′], k' is the total number of features in the second covariance matrix, p'≤k'; α′ P λ represents the cumulative contribution rate of the first p' features in the second covariance matrix. i ′ represents the eigenvalue of the i'th feature in the second covariance matrix.

[0039] In some embodiments, the formula for calculating the upper limit value of each of the second key early warning indicators includes:

[0040] The calculation formulas for the lower limits of each of the second key early warning indicators include:

[0041] In the above formula, Y up Y represents the upper limit of each of the second key early warning indicators. low These are the lower limits for each of the second key early warning indicators. Var is the mean of the second new feature data after clustering corresponding to each second key early warning indicator. * (y) represents the sample variance corresponding to each of the second key early warning indicators, Z 1-α / 2 α is the critical value corresponding to 100·(1-α)% of the standard normal distribution, where α is the adjustment coefficient.

[0042] In some embodiments, the distribution station area is a distribution station area that includes an electric vehicle charging station.

[0043] According to a second aspect of the embodiments of this application, a risk classification and early warning system for distribution radio areas is provided, comprising:

[0044] An extraction unit is used to extract features from the target monitoring data of the distribution station area to obtain the first data features of the distribution station area; the target monitoring data of the distribution station area is collected based on a first data acquisition time interval.

[0045] The first acquisition unit is used to perform correlation analysis on preset feature indicators based on the data characteristics of the first transformer area to obtain the first key early warning indicator.

[0046] The second acquisition unit is used to fuse feature indicators using the first key early warning indicator to obtain the first new feature data.

[0047] The third acquisition unit is used to predict the warning level of the distribution radio station area based on the first new feature data and using the warning model.

[0048] The adjustment unit is used to adjust the first data acquisition time interval according to the warning level of the distribution radio area to obtain the second data acquisition time interval;

[0049] The fourth acquisition unit is used to reacquire the target monitoring data based on the second data acquisition time interval, and to redetermine the warning level of the distribution radio station area based on the reacquired target monitoring data.

[0050] In some embodiments, the extraction unit includes:

[0051] The data cleaning subunit is used to clean the target monitoring data.

[0052] The first acquisition subunit is used to extract data corresponding to preset feature indicators from the cleaned target monitoring data to obtain the data features of the first transformer area.

[0053] In some embodiments, the first acquisition unit includes:

[0054] The second acquisition subunit is used to perform correlation analysis on the preset feature indicators based on the data characteristics of the first transformer area using a grayscale correlation analysis method to obtain the first key early warning indicator.

[0055] In some embodiments, the second acquisition subunit includes:

[0056] The first construction module is used to construct a first comparison sequence using the data features of the first transformer area;

[0057] The second construction module is used to construct a first reference sequence using the collected first historical monitoring data;

[0058] The processing module is used to standardize both the first comparison sequence and the first reference sequence;

[0059] The first calculation module is used to determine the absolute difference between the first comparison sequence after standardization of each preset feature index and the first reference sequence after standardization of the feature index for each preset feature index.

[0060] The determination module is used to determine the gray-level correlation between the first comparison sequence after standardization of each preset feature indicator and the first reference sequence after standardization of the feature indicator, using the absolute difference corresponding to the feature indicator for each preset feature indicator.

[0061] The selection module is used to select a preset feature index corresponding to the grayscale correlation degree that meets preset conditions as the first key early warning index.

[0062] In some embodiments, the second building module is further configured to:

[0063] Perform data cleaning on the first historical monitoring data;

[0064] Standardize the first historical monitoring data after data cleaning corresponding to the preset feature indicators to obtain the first historical monitoring data after standardization corresponding to the preset feature indicators.

[0065] Let the minimum value in the first historical monitoring data after standardization corresponding to each preset feature index be the optimal value, and the maximum value in the first historical monitoring data after standardization corresponding to each preset feature index be the worst value.

[0066] The first reference sequence is constructed using the optimal and worst values ​​corresponding to the preset feature indicators.

[0067] In some embodiments, the determining module is further configured to:

[0068] Based on the absolute difference, determine the maximum absolute difference and minimum absolute difference between the first comparison series after standardization corresponding to the preset feature index and the first reference series after standardization corresponding to the preset feature index.

[0069] Using the maximum and minimum absolute differences, the gray-scale correlation between the first comparison series and the first reference series after standardization, corresponding to each preset feature index, is determined.

[0070] In some embodiments, the second acquisition unit includes:

[0071] The first construction subunit is used to extract data corresponding to the first key early warning indicator from the target monitoring data after data cleaning, and construct the first feature set matrix.

[0072] The second construction subunit is used to construct the first new feature data based on the first feature set matrix using principal component analysis.

[0073] In some embodiments, the second building subunit includes:

[0074] The second calculation module is used to determine the first covariance matrix using the first feature set matrix;

[0075] The third calculation module is used to determine the cumulative contribution rate of the first p features in the first covariance matrix;

[0076] The third construction module is used to construct the first new feature data using the first p features in the first covariance matrix when the cumulative contribution rate of the first p features is greater than or equal to the expected value of the cumulative contribution rate.

[0077] Where p is a positive integer, p≤k, and k is the total number of features in the first covariance matrix.

[0078] In some embodiments, the adjustment unit is specifically used for:

[0079] When the warning level is the fourth warning level or the number of electric vehicles entering the charging station in the distribution area reaches the first percentage of the charging station capacity, the second data collection time interval is the first multiple of the first data collection time interval.

[0080] When the warning level is the third warning level or the number of electric vehicles entering the charging station in the distribution area reaches the second percentage of the charging station capacity, the second data collection time interval is the second multiple of the first data collection time interval.

[0081] When the warning level is lower than or equal to the second warning level, and the number of electric vehicles entering the distribution area is less than the second percentage of the charging station capacity, the second data collection time interval is a third multiple of the first data collection time interval.

[0082] When the warning level is lower than or equal to the first warning level, and the number of electric vehicles entering the distribution area is less than the second percentage of the charging station capacity, the second data collection time interval is the first data collection time interval.

[0083] In some embodiments, the distribution station area is a distribution station area that includes an electric vehicle charging station.

[0084] According to a third aspect of the present application, an electronic device is provided, comprising: at least one processor and a memory; wherein the memory and the processor are connected via a bus.

[0085] The memory is used to store one or more programs;

[0086] When the one or more programs are executed by the at least one processor, the aforementioned method for risk classification and early warning of distribution radio areas is implemented.

[0087] According to a fourth aspect of the embodiments of this application, a readable storage medium is provided, on which an executable program is stored, wherein when the executable program is executed, the aforementioned distribution area risk classification and early warning method is implemented.

[0088] According to a fifth aspect of the embodiments of this application, a computer program product is provided, the computer program product including a computer program or instructions, which, when the computer program or instructions are executed on a computer, cause the computer to implement the aforementioned distribution area risk classification and early warning method.

[0089] The technical solution provided in this application has the following beneficial effects:

[0090] This application provides a method, system, device, storage medium, and program product for risk classification and early warning of distribution transformer substations. It extracts features from target monitoring data of the substation collected at a first data acquisition time interval to obtain first substation data features. Based on these first substation data features, it performs correlation analysis on preset feature indicators to obtain first key early warning indicators. It then fuses feature indicators using these first key early warning indicators to obtain first new feature data. Based on this new feature data, it uses an early warning model to predict the early warning level of the substation. Based on the early warning level, it adjusts the first data acquisition time interval to obtain a second data acquisition time interval. Finally, it re-acquires target monitoring data based on the second data acquisition time interval and re-obtains the early warning level of the substation based on the re-acquired target monitoring data. This not only enables flexible adjustment of the acquisition interval for the substation, reducing the pressure on data acquisition resources and the workload of data transmission and retrieval, but also improves the accuracy and reliability of risk early warning for distribution transformer substations, enhances the convenience of electricity services for transportation in substations containing electric vehicle charging stations, and ultimately improves the safety and stability of the distribution network system. Attached Figure Description

[0091] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0092] Figure 1 is a flowchart of a risk classification and early warning method for a distribution area provided in an embodiment of this application;

[0093] Figure 2 is a structural block diagram of a risk classification and early warning system for a distribution area provided in an embodiment of this application;

[0094] Figure 3 is a structural block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0095] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the following embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0096] This application provides a method for risk classification and early warning of power distribution areas, as shown in Figure 1, including the following steps:

[0097] Step 11: Extract features from the target monitoring data of the distribution station area to obtain the first data feature of the distribution station area; the target monitoring data of the distribution station area is collected based on the first data acquisition time interval;

[0098] Step 12: Based on the data characteristics of the first transformer area, perform correlation analysis on the preset feature indicators to obtain the first key early warning indicator;

[0099] Step 13: Use the first key early warning indicator to fuse feature indicators and obtain the first new feature data;

[0100] Step 14: Based on the first new feature data, use the early warning model to predict the early warning level of the distribution radio area;

[0101] Step 15: Adjust the first data acquisition time interval according to the warning level of the distribution radio area to obtain the second data acquisition time interval;

[0102] Step 16: Reacquire target monitoring data based on the second data acquisition time interval, and re-obtain the warning level of the distribution radio station area based on the reacquired target monitoring data.

[0103] In some embodiments, after obtaining the warning information, i.e. the warning level of the distribution radio area, the warning information can be stored as a reference for handling.

[0104] In some embodiments, the distribution radio area is a distribution radio area that includes electric vehicle charging stations.

[0105] This application enables more efficient monitoring and more accurate early warning of the operating status of distribution transformer areas containing electric vehicle charging stations, thereby improving the energy supply level and reliability of the transformer areas where charging stations are located.

[0106] In some embodiments, step 11 includes:

[0107] Step 111: Perform data cleaning on the target monitoring data;

[0108] Step 112: Extract data corresponding to preset feature indicators from the cleaned target monitoring data to obtain the data features of the first monitoring area.

[0109] In some embodiments, step 12 includes:

[0110] Step 121: Based on the data characteristics of the first transformer area, the gray-scale correlation analysis method is used to perform correlation analysis on the preset feature indicators to obtain the first key early warning indicator.

[0111] In some embodiments, the first key warning indicator may include, but is not limited to,: heavy overload of the transformer area, three-phase voltage imbalance, voltage deviation, and frequency deviation.

[0112] In some embodiments, step 121 includes:

[0113] Step 1211: Construct the first comparison sequence using the data features of the first transformer area;

[0114] Step 1212: Construct a first reference sequence using the collected first historical monitoring data;

[0115] Step 1213: Standardize both the first comparison sequence and the first reference sequence;

[0116] Step 1214: Calculate the absolute difference between the first standardized comparison series and the first standardized reference series corresponding to each preset feature index;

[0117] Step 1215: Using the absolute difference, determine the gray-level correlation between the first comparison series and the first reference series after standardization for each preset feature index;

[0118] Step 1216: Select the preset feature index corresponding to the gray-scale correlation degree that meets the preset conditions as the first key early warning index.

[0119] It should be noted that this application does not limit the "preset conditions," which can be set by those skilled in the art based on experimental data, expert experience, or engineering needs. In some embodiments, the preset conditions may include, but are not limited to, arranging the grayscale correlation in descending order and selecting the feature index with the highest grayscale correlation as the first key early warning index.

[0120] In some embodiments, step 1212 includes:

[0121] Step 1212a: Clean the first historical monitoring data;

[0122] Step 1212b: Standardize the first historical monitoring data after data cleaning corresponding to the preset feature indicators to obtain the first historical monitoring data after standardization corresponding to the preset feature indicators.

[0123] Step 1212c: Let the minimum value in the first historical monitoring data after standardization corresponding to each preset feature index be the optimal value, and the maximum value in the first historical monitoring data after standardization corresponding to each preset feature index be the worst value.

[0124] Step 1212d: Construct the first reference sequence using the optimal and worst values ​​corresponding to the preset feature indicators.

[0125] In some embodiments, step 1215 includes:

[0126] Step 1215a: Based on the absolute difference, determine the maximum and minimum absolute difference between the first standardized comparison series and the first standardized reference series corresponding to each preset feature index;

[0127] Step 1215b: Using the maximum absolute difference and the minimum absolute difference, calculate the gray-scale correlation degree between the first comparison series and the first reference series after standardization for each preset feature index.

[0128] In some embodiments, step 13 includes:

[0129] Step 131: Extract the data corresponding to the first key early warning indicator from the cleaned target monitoring data and construct the first feature set matrix;

[0130] Step 132: Based on the first feature set matrix, construct the first new feature data using principal component analysis.

[0131] In some embodiments, step 132 includes:

[0132] Step 1321: Calculate the first covariance matrix using the first feature set matrix;

[0133] Specifically, the formula for calculating the first covariance matrix includes: B = X T X

[0134] In the above formula, X is the first characteristic set matrix, and T is the transpose;

[0135] Step 1322: Calculate the cumulative contribution rate of the first p features in the first covariance matrix;

[0136] Specifically, the formula for calculating the cumulative contribution rate of the first p features in the first covariance matrix includes:

[0137] In the above formula, B is the first covariance matrix; i∈[1,k], k is the total number of features in the first covariance matrix, p≤k; α Pλ represents the cumulative contribution rate of the first p features in the first covariance matrix. i Let be the eigenvalue of the i-th feature in the first covariance matrix;

[0138] Step 1323: When the cumulative contribution rate of the current p features is greater than or equal to the expected value of the cumulative contribution rate, construct the first new feature data using the first p features in the first covariance matrix;

[0139] Where p is a positive integer, p≤k, and k is the total number of features in the first covariance matrix.

[0140] In some embodiments, step 15 includes:

[0141] Step 151: If the warning level is the fourth warning level or the number of electric vehicles entering the distribution area reaches the first percentage of the charging station capacity, then the second data collection time interval is the first multiple of the first data collection time interval.

[0142] Step 152: If the warning level is the third warning level or the number of electric vehicles entering the distribution area reaches the second percentage of the charging station capacity, then the second data collection time interval is the second multiple of the first data collection time interval;

[0143] Step 153: If the warning level is lower than or equal to the second warning level, and the number of electric vehicles entering the distribution area is less than the second percentage of the charging station capacity, then the second data collection time interval is the third multiple of the first data collection time interval.

[0144] Step 154: If the warning level is lower than or equal to the first warning level, and the number of electric vehicles entering the distribution area is less than the second percentage of the charging station capacity, then the second data collection time interval is the first data collection time interval.

[0145] For example, let the time interval for collecting target monitoring data be t0, the first warning level be green, the second warning level be blue, the third warning level be orange, and the fourth warning level be red;

[0146] If the warning level reaches the red warning level, or the number of electric vehicles entering the site reaches 90% of the charging station's capacity, the data collection time interval for the target monitoring will be adjusted to t0 / 10.

[0147] If the warning level reaches the orange warning level, or the number of electric vehicles entering the site reaches 80% of the charging station's capacity, the time interval for collecting target monitoring data will be adjusted to t0 / 8.

[0148] If the warning levels are all no higher than the blue warning level, and the number of electric vehicles entering the site is less than 80% of the charging station's capacity, then the data collection time interval for the target monitoring will be adjusted to t0 / 5.

[0149] If the warning levels are all no higher than the green warning level, and the number of electric vehicles entering the site is less than 80% of the charging station capacity, then the time interval for collecting target monitoring data will be maintained at t0.

[0150] The risk classification and early warning method for distribution transformer substations based on data monitoring time interval adjustment provided in this application analyzes the operating status and provides early warnings based on monitoring data and historical data of distribution transformer substations containing electric vehicle charging stations. It flexibly adjusts the data collection time interval of the substations according to the early warning results and the data collection time adjustment criteria, thereby realizing flexible monitoring and classification early warning under different operating conditions of distribution transformer substations containing electric vehicle charging stations.

[0151] In some embodiments, the method further includes: step 10: establishing an early warning model;

[0152] Step 10 includes:

[0153] Step 101: Collect the second historical monitoring data of the distribution radio area;

[0154] Step 102: Extract features from the second historical monitoring data to obtain the features of the second transformer area data;

[0155] Step 103: Based on the data characteristics of the second transformer area, perform correlation analysis on the preset feature indicators to obtain the second key early warning indicator;

[0156] Step 104: Use the second key early warning indicator to fuse feature indicators and obtain the second new feature data;

[0157] Step 105: Determine the historical early warning level of the distribution radio station corresponding to the second new feature data;

[0158] Step 106: Construct a dataset using the second new feature data and the historical warning levels of the distribution radio area, and use the dataset to train the long and short neural network model to obtain the warning model.

[0159] In some embodiments, step 102 includes:

[0160] Step 1021: Clean the second set of historical monitoring data;

[0161] Step 1022: Extract data corresponding to preset feature indicators from the second historical monitoring data after data cleaning to obtain the data features of the second transformer area.

[0162] In some embodiments, step 103 includes:

[0163] Step 1031: Based on the data characteristics of the second transformer area, the gray-scale correlation analysis method is used to perform correlation analysis on the preset feature indicators to obtain the second key early warning indicator.

[0164] In some embodiments, step 1031 includes:

[0165] Step 1031a: Construct a second comparison sequence using the data features of the second transformer area;

[0166] Step 1031b: Construct a second reference series using the collected third historical monitoring data;

[0167] Step 1031c: Standardize both the second comparison sequence and the second reference sequence;

[0168] Step 1031d: Calculate the absolute difference between the standardized second comparison series and the standardized second reference series corresponding to each preset feature index;

[0169] Step 1031e: Using the absolute difference, determine the gray-level correlation between the standardized second comparison series and the standardized second reference series corresponding to each preset feature index;

[0170] Step 1031f: Select the feature index corresponding to the gray-scale correlation degree that meets the preset conditions as the second key early warning index;

[0171] The third historical monitoring data was collected earlier than the second historical monitoring data.

[0172] It should be noted that this application does not limit the "preset conditions," which can be set by those skilled in the art based on experimental data, expert experience, or engineering needs. In some embodiments, the preset conditions may include, but are not limited to, arranging the grayscale correlation in descending order and selecting the feature index with the highest grayscale correlation as the second key early warning index.

[0173] In some embodiments, step 1031b includes:

[0174] Data cleaning was performed on the third historical monitoring data;

[0175] Standardize the third historical monitoring data after cleaning the data corresponding to the preset feature indicators to obtain the standardized third historical monitoring data corresponding to the preset feature indicators.

[0176] Let the minimum value in the standardized third historical monitoring data corresponding to each preset feature indicator be the optimal value, and the maximum value in the standardized third historical monitoring data corresponding to each preset feature indicator be the worst value.

[0177] A second reference sequence is constructed using the optimal and worst values ​​corresponding to the preset feature indicators.

[0178] In some embodiments, step 1031e includes:

[0179] Based on the absolute difference, determine the maximum and minimum absolute difference between the standardized second comparison series and the standardized second reference series corresponding to each preset feature index;

[0180] Using the maximum and minimum absolute differences, the gray-scale correlation degree between the standardized second comparison series and the standardized second reference series corresponding to each preset feature index is calculated.

[0181] In some embodiments, step 104 includes:

[0182] Step 1041: Extract the data corresponding to the second key early warning indicator from the second historical monitoring data after data cleaning, and construct the second feature set matrix;

[0183] Step 1042: Based on the second feature set matrix, construct the second new feature data using principal component analysis.

[0184] In some embodiments, step 1042 includes:

[0185] Step 1042a: Calculate the second covariance matrix using the second feature set matrix;

[0186] Specifically, the formula for calculating the second covariance matrix includes: B' = X' T X'

[0187] In the above formula, X' is the second characteristic set matrix, and T is the transpose;

[0188] Step 1042b: Calculate the cumulative contribution rate of the first p' features in the second covariance matrix;

[0189] Specifically, the formula for calculating the cumulative contribution rate of the first p' features in the second covariance matrix includes:

[0190] In the above formula, X' is the second feature set matrix, T is the transpose, B' is the second covariance matrix; i'∈[1,k'], k' is the total number of features in the second covariance matrix, p'≤k'; α′ P λ represents the cumulative contribution rate of the first p' features in the second covariance matrix. i ' represents the eigenvalue of the i'th feature in the second covariance matrix;

[0191] Step 1042c: When the cumulative contribution rate of the current p' features is greater than or equal to the expected value of the cumulative contribution rate, construct the second new feature data using the first p' features in the second covariance matrix;

[0192] Where p' is a positive integer, p'≤k', and k' is the total number of features in the second covariance matrix.

[0193] In some embodiments, step 105 includes:

[0194] Step 1051: Based on the preset standard limit values ​​for normal operation and the preset standard limit values ​​for fault operation of the transformer area corresponding to the second key early warning indicator, extract the effective data from the second new feature data to obtain the processed second new feature data.

[0195] Step 1052: Based on the second key early warning indicator, use the K-means clustering method to perform feature clustering on the processed second new feature data to obtain the clustered second new feature data;

[0196] Step 1053: Based on the second new feature data after clustering, the interval prediction method is used to analyze and obtain the extreme values ​​of change corresponding to each second key early warning indicator;

[0197] For example, assuming the key early warning indicators are transformer area overload, three-phase voltage imbalance, voltage deviation, and frequency deviation, the effective data in the second new feature data can be extracted based on the national standard limits of the transformer area during normal operation and fault operation corresponding to these four key early warning indicators. The processed second new feature data is then obtained, and the K-means clustering method is used to perform feature clustering on the early warning objects (i.e., transformer area overload, three-phase voltage imbalance, voltage deviation, and frequency deviation). The changes in the indicators and their extreme values ​​are analyzed through interval prediction.

[0198] Step 1054: Based on the clustered second new feature data and the extreme values ​​of change corresponding to each second key early warning indicator, use the Bootstrap algorithm to predict the sample variance of the Bootstrap sample corresponding to each second key early warning indicator.

[0199] Step 1055: Calculate the upper and lower limits of each second key early warning indicator using the sample variance corresponding to each second key early warning indicator;

[0200] Specifically, the calculation formulas for the upper limits of each of the second key early warning indicators include:

[0201] The calculation formulas for the lower limits of each of the second key early warning indicators include:

[0202] In the above formula, Y upY represents the upper limit of each of the second key early warning indicators. low These are the lower limits for each of the second key early warning indicators. Var represents the mean of the second new feature data after clustering for each of the second key early warning indicators. * (y) represents the sample variance corresponding to each of the second key early warning indicators, Z 1-α / 2 α is the critical value corresponding to 100·(1-α)% of the standard normal distribution, where α is the adjustment coefficient.

[0203] Step 1056: Determine the historical warning level of the distribution area corresponding to the second new feature data based on the upper and lower limits of each second key warning indicator.

[0204] It should be noted that the method of "predicting the sample variance of the Bootstrap samples corresponding to each second key early warning indicator using the Bootstrap algorithm" involved in the embodiments of this application is well known to those skilled in the art; therefore, its specific implementation will not be described in detail. In some embodiments, the formula for calculating the sample variance includes:

[0205] In the above formula, b∈[1,B], and B is the total number of Bootstrap samples; Var * (y) represents the sample variance. Let be the b-th Bootstrap sample, and let ΔY be the sample mean.

[0206] In some embodiments, step 1056 includes:

[0207] Step 1056a: When the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the first confidence threshold, the historical early warning level of the distribution area is the first early warning level;

[0208] Step 1056b: When the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the second confidence threshold, the historical early warning level of the distribution area is the second early warning level;

[0209] Step 1056c: When the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the third confidence threshold, the historical early warning level of the distribution area is the third early warning level.

[0210] Step 1056d: When the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the fourth confidence threshold, the historical early warning level of the distribution area is the fourth early warning level.

[0211] It should be noted that this application does not limit the "first confidence threshold, second confidence threshold, third confidence threshold and fourth confidence threshold", which can be set by those skilled in the art based on experimental data, expert experience or engineering needs.

[0212] For example, the 98% confidence fluctuation range is defined as Level 1 (i.e., the first warning level), the 95% confidence fluctuation range is defined as Level 2 (i.e., the second warning level), the 90% confidence fluctuation range is defined as Level 3 (i.e., the third warning level), and the 86% confidence fluctuation range is defined as Level 4 (i.e., the fourth warning level). Then, according to the severity of the risk warning, it is divided into four warning levels from low to high: green (Level 1), blue (Level 2), orange (Level 3), and red (Level 4), thus completing the warning level classification.

[0213] In some embodiments, step 106 involves training a long and short neural network model using a dataset to obtain an early warning model, including:

[0214] Step 1061: Divide the dataset into a training set, a validation set, and a test set;

[0215] Step 1062: Train the long and short neural network model using the training set to obtain the trained long and short neural network model;

[0216] Step 1063: Validate the trained long and short neural network model using the validation set. If the validation is successful, evaluate the performance of the trained long and short neural network model using the test set. If the validation fails, adjust the hyperparameters of the long and short neural network model, monitor for overfitting, and retrain the long and short neural network model until the validation is successful.

[0217] Step 1064: Use the test set to evaluate the performance of the trained long and short neural network model. If the performance evaluation of the trained long and short neural network model is successful, then the trained long and short neural network model is an early warning model; if the performance evaluation of the trained long and short neural network model fails, then the trained long and short neural network model does not contain the features of the transformer area.

[0218] In some embodiments, statistical operating data from nearly three years is divided into training, validation, and test sets by year. The data from the first year is used as the training set to train the neural network model; the data from the second year is used as the validation set to adjust the model's hyperparameters and monitor overfitting; the data from the third year is used as the test set to evaluate the model's performance; based on the aforementioned new features as input features for historical data, the system is trained on the effective historical data to output the risk prediction results for each indicator in the next period, thereby determining the charging safety risks and distribution network risks; when the warning value is greater than the warning boundary value, a warning is issued; if it is less than the boundary value, the system is in normal operating condition and no warning is issued, thus completing the hierarchical warning system.

[0219] To further illustrate the above-mentioned risk classification and early warning method for distribution radio areas, this application provides a specific example, including the following steps:

[0220] S100: For the distribution transformer areas requiring monitoring and early warning, data is collected from the monitored and early warning areas using an initial time interval t0 as the data acquisition time interval. The target monitoring data may include, but is not limited to, the following: voltage, current, and power of each distribution area; equipment status (e.g., whether the transformer is operating normally); fault recordings; electric vehicle charging power; and the number of connected electric vehicles N. ev Data such as electric vehicle charging stations are included in the distribution radio areas that require monitoring and early warning.

[0221] S200: Preprocess the target monitoring data and extract the data features of the first monitoring area;

[0222] Specifically, the preprocessing of target monitoring data includes: using generative adversarial networks to fill in missing data by sequentially constructing generator and discriminator networks, defining the data completion loss function, training the adversarial network, generating missing values, and evaluating the consistency of generated values.

[0223] The data feature extraction for the first transformer area includes, but is not limited to: overall shape, fluctuation, trend, correlation, date type, voltage deviation, power factor, three-phase unbalance and harmonic current, etc.

[0224] In some embodiments, for the overall shape of the data of the transformer area, the daily and monthly electricity consumption rate is used to reflect the overall change in electricity consumption, the peak-valley difference rate is used to reflect the magnitude of the change in electricity consumption, and the quarterly electricity consumption ratio is used to reflect the distribution of electricity consumption.

[0225] To address the power fluctuations in charging stations, a coefficient of variation is used to represent the degree of deviation of the power consumption curve of the charging station from the average value curve, in order to reflect the fluctuations in the power consumption of the charging station. The difference between the first and last monthly power consumption data is used to reflect the overall fluctuations in the power consumption of the charging station.

[0226] Based on the trend characteristics of the data in the distribution area m t The daily electricity consumption sequence of the charging station area is calculated using the simple moving average method and the following formula:

[0227] In the above formula, n is the number of data points within the time period, t is the time point, and m is the number of data points within the time period. t-1 Let m represent the trend characteristics of the transformer area data at time t-1. t-2 Let m represent the trend characteristics of the transformer area data at time t-2. t-n S represents the trend characteristics of the transformer area data at time tn; tLet t be the daily electricity consumption of the charging station area at time t;

[0228] Compare the sequence M of each charging station area at time t one by one. t and sequence S t M t Greater than S t The points are denoted as: {a1, a2, ..., a e ,…,a u}, that is, record the data points where the actual value is greater than the predicted value, where a1 is the first M t Greater than S t Point a2 is the second M t Greater than S t point a e For the e-th M t Greater than S t point a u For the u-th M t Greater than S t The point; M t Less than S t The points are denoted as: {b1, b2, ..., b w ,…,b v}, that is, record the data points where the actual value is smaller than the predicted value, where b1 is the first M. t Less than S t Point b2 is the second M t Less than S t point b w For the w-th M t Less than S t point b v For the vth M t Less than S t The points; then, the upward trend tra and the downward trend trb are calculated according to the following formulas:

[0229] In the above formula, u is M t Greater than S t The total number of points; v is M t Less than S t The total number of points.

[0230] S300: Analyze the correlation between the above feature indicators and select key features of the transformer area data for feature indicator fusion:

[0231] The correlation degree calculation adopts the gray-scale correlation analysis method. A first comparison series is constructed using the data characteristics of the first monitoring area, and a first reference series is constructed using the collected second historical monitoring data. Both the first comparison series and the first reference series are standardized to eliminate dimensional differences. The absolute difference between the standardized first comparison series and the standardized first reference series corresponding to each preset feature index is calculated, and the maximum and minimum absolute differences are listed to obtain the maximum and minimum absolute differences. Using the maximum and minimum absolute differences, the gray-scale correlation degree between the standardized first comparison series and the standardized first reference series corresponding to each preset feature index is calculated. The correlation coefficient between the series can be obtained using the average value calculation method.

[0232] The grayscale correlation is sorted in descending order, and the feature index with the highest correlation is selected as the first key early warning index. The first key early warning index includes: transformer area overload, three-phase voltage imbalance, voltage deviation, and frequency deviation, and is used for graded early warning.

[0233] Let the first feature set matrix composed of the first key early warning indicators be X (it can be understood that each row of the feature set matrix is ​​a record), and use the principal component analysis method to construct the first new feature data;

[0234] In some embodiments, constructing the first new feature data using principal component analysis includes: constructing a centered matrix with column mean of 0, and calculating the eigenvalues ​​{λ1,λ2,λ3,...,λ} of the covariance matrix B according to the following three formulas. k} and find its corresponding orthonormal vectors {y1,y2,y3,...,y}. k}: B=X T X = VΛV T V = (y1, y2, ..., y k ) Λ=(λ1,λ2,...,λ k )

[0235] In the above formula, λ1 is the first eigenvalue, λ2 is the second eigenvalue, λ3 is the third eigenvalue, and λ... k Let y1 be the k-th eigenvalue, V be the orthonormal vector of the covariance matrix B, T be the transpose, Λ be the eigenvalue of the covariance matrix B, y1 be the orthonormal vector of λ1, y2 be the orthonormal vector of λ2, y3 be the orthonormal vector of λ3, and y4 be the orthonormal vector of λ3. k For λ k Orthogonal vectors;

[0236] Calculate the cumulative contribution rate of the first P new features according to the following formula, set the expected value of the cumulative contribution rate as a0, and let a p >a0 determines the number of new features, resulting in new feature U.j =Xy j j = 1, 2, ..., p:

[0237] In the above formula, i∈[1,k], k is the total number of features in the first covariance matrix, p≤k; α P λ represents the cumulative contribution rate of the first p features in the first covariance matrix. i y represents the eigenvalue of the i-th feature in the first covariance matrix; j for.

[0238] S400: Using the first new feature data as input to the early warning model, the output is the predicted early warning level of the distribution radio area;

[0239] S500: Combining the warning levels of the four primary key warning indicators—overload of transformer area, three-phase voltage imbalance, voltage deviation, and frequency deviation—with the number of electric vehicles entering the site, the data collection time interval t0 is adjusted based on the data collection time interval adjustment criteria.

[0240] The criteria for adjusting the data acquisition time interval are as follows:

[0241] 1) If one of the four key early warning indicators reaches the red warning level, or the number of electric vehicles entering the site reaches 90% of the charging station capacity, the data collection time interval will be adjusted to t0 / 10.

[0242] 2) If one of the four key early warning indicators reaches the orange warning level, or the number of electric vehicles entering the site reaches 80% of the charging station capacity, the data collection time interval will be adjusted to t0 / 8.

[0243] 3) If all four key early warning indicators are not higher than the blue warning level, and the number of electric vehicles entering the site is less than 80% of the charging station capacity, then the data collection time interval will be adjusted to t0 / 5.

[0244] 4) If none of the four primary key early warning indicators are higher than the green early warning level, and the number of electric vehicles entering the site is less than 80% of the charging station capacity, then the data collection time interval will be maintained at t0.

[0245] S600: The updated data acquisition time interval is sent to the data acquisition terminal in the distribution area to adjust the data acquisition frequency of the distribution area, and the early warning information is stored and used as a reference for handling.

[0246] This application provides a risk classification and early warning method for distribution transformer substations. It employs data feature analysis and feature index correlation analysis to calculate the operational status classification and early warning thresholds based on substation operation monitoring data and historical data, enabling a more detailed evaluation of the substation's operational status. This application proposes a criterion based on operational status analysis and the data collection time adjustment for the number of electric vehicles, allowing for flexible adjustment of the data collection interval for distribution transformer substations. This reduces the pressure on data collection resources and the workload of data transmission and retrieval, improving the accuracy and reliability of risk early warning for distribution transformer substations. Furthermore, the inclusion of electric vehicle charging stations in this application enhances the convenience of electricity services for transportation, leading to increasing engineering practice. This application can be applied to the field of distribution network operation monitoring, improving the safety and stability of distribution network systems.

[0247] This application also provides a risk classification and early warning system for distribution radio areas, as shown in Figure 2, including:

[0248] Extraction unit 201 is used to extract features from the target monitoring data of the distribution station area collected based on the first data acquisition time interval to obtain the data features of the first distribution station area;

[0249] The first acquisition unit 202 is used to perform correlation analysis on preset feature indicators based on the data characteristics of the first transformer area to obtain the first key early warning indicator;

[0250] The second acquisition unit 203 is used to fuse feature indicators using the first key early warning indicator to obtain the first new feature data;

[0251] The third acquisition unit 204 is used to predict the warning level of the distribution radio station area based on the first new feature data and using the warning model.

[0252] The adjustment unit 205 is used to adjust the first data acquisition time interval according to the warning level of the distribution radio area to obtain the second data acquisition time interval;

[0253] The fourth acquisition unit 206 is used to reacquire target monitoring data based on the second data acquisition time interval, and reacquire the warning level of the distribution radio station area based on the reacquired target monitoring data.

[0254] In some embodiments, the extraction unit 201 includes:

[0255] The data cleaning subunit is used to clean the target monitoring data.

[0256] The first acquisition subunit is used to extract data corresponding to preset feature indicators from the cleaned target monitoring data to obtain the data features of the first transformer area.

[0257] In some embodiments, the first acquisition unit 202 includes:

[0258] The second acquisition subunit is used to perform correlation analysis on preset feature indicators based on the data characteristics of the first transformer area, using gray-scale correlation analysis method to obtain the first key early warning indicator.

[0259] In some embodiments, the second acquisition subunit includes:

[0260] The first construction module is used to construct the first comparison sequence using the data characteristics of the first transformer area;

[0261] The second construction module is used to construct a first reference sequence using the collected first historical monitoring data;

[0262] The processing module is used to standardize both the first comparison sequence and the first reference sequence;

[0263] The first calculation module is used to calculate the absolute difference between the first comparison series and the first reference series after standardization for each preset feature index.

[0264] The determination module is used to determine the gray-scale correlation between the first comparison series and the first reference series after standardization, corresponding to each preset feature index, using the absolute difference.

[0265] The selection module is used to select the preset feature index corresponding to the gray-scale correlation degree that meets the preset conditions as the first key early warning index.

[0266] In some embodiments, the second building module is specifically used for:

[0267] Data cleaning was performed on the first historical monitoring data;

[0268] Standardize the first historical monitoring data after cleaning the data corresponding to the preset feature indicators to obtain the first historical monitoring data after standardization corresponding to the preset feature indicators.

[0269] Let the minimum value in the first historical monitoring data after standardization corresponding to each preset feature index be the optimal value, and the maximum value in the first historical monitoring data after standardization corresponding to each preset feature index be the worst value.

[0270] The first reference sequence is constructed using the optimal and worst values ​​corresponding to the preset feature indicators.

[0271] In some embodiments, the determining module is specifically used for:

[0272] Based on the absolute difference, determine the maximum and minimum absolute difference between the first standardized comparison series and the first standardized reference series corresponding to each preset feature index;

[0273] Using the maximum and minimum absolute differences, the gray-level correlation between the first comparison series and the first reference series after standardization for each preset feature index is calculated.

[0274] In some embodiments, the second acquisition unit 203 includes:

[0275] The first construction subunit is used to extract data corresponding to the first key early warning indicator from the cleaned target monitoring data and construct the first feature set matrix.

[0276] The second construction subunit is used to construct the first new feature data based on the first feature set matrix using the principal component analysis method.

[0277] In some embodiments, the second building subunit includes:

[0278] The second calculation module is used to calculate the first covariance matrix using the first feature set matrix;

[0279] The third calculation module is used to calculate the cumulative contribution rate of the first p features in the first covariance matrix;

[0280] The third construction module is used to construct the first new feature data using the first p features in the first covariance matrix when the cumulative contribution rate of the current p features is greater than or equal to the expected value of the cumulative contribution rate.

[0281] Where p is a positive integer, p≤k, and k is the total number of features in the first covariance matrix.

[0282] In some embodiments, the adjusting unit 205 is specifically used for:

[0283] If the warning level is the fourth warning level or the number of electric vehicles entering the distribution area reaches the first percentage of the charging station capacity, then the second data collection time interval is the first multiple of the first data collection time interval.

[0284] If the warning level is the third warning level or the number of electric vehicles entering the distribution area reaches the second percentage of the charging station capacity, then the second data collection time interval is the second multiple of the first data collection time interval.

[0285] If the warning level is lower than or equal to the second warning level, and the number of electric vehicles entering the distribution area is less than the second percentage of the charging station capacity, then the second data collection time interval is the third multiple of the first data collection time interval.

[0286] If the warning level is lower than or equal to the first warning level, and the number of electric vehicles entering the distribution area is less than the second percentage of the charging station capacity, then the second data collection time interval is the first data collection time interval.

[0287] In some embodiments, the apparatus further includes: a modeling unit for establishing an early warning model; the modeling unit includes:

[0288] The acquisition subunit is used to collect the second historical monitoring data of the distribution radio station area;

[0289] The third acquisition subunit is used to extract features from the second historical monitoring data to obtain the features of the second transformer area data;

[0290] The fourth acquisition subunit is used to perform correlation analysis on preset feature indicators based on the data characteristics of the second transformer area to obtain the second key early warning indicator;

[0291] The fifth acquisition subunit is used to fuse feature indicators using the second key early warning indicator to obtain the second new feature data;

[0292] The sub-unit is determined to identify the historical early warning level of the distribution radio station corresponding to the second new feature data.

[0293] The sixth acquisition subunit is used to construct a dataset using the second new feature data and the historical warning levels of the distribution radio area, and to train the long and short neural network model using the dataset to obtain the warning model.

[0294] In some embodiments, the third acquisition subunit includes:

[0295] The data cleaning module is used to clean the second historical monitoring data;

[0296] The first acquisition module is used to extract data corresponding to preset feature indicators from the second historical monitoring data after data cleaning, so as to obtain the data features of the second transformer area.

[0297] In some embodiments, the fourth acquisition subunit includes:

[0298] The second acquisition module is used to perform correlation analysis on preset feature indicators based on the data characteristics of the second transformer area, and obtain the second key early warning indicator.

[0299] In some embodiments, the second acquisition module includes:

[0300] The first construction submodule is used to construct the second comparison sequence using the data characteristics of the second transformer area;

[0301] The second construction submodule is used to construct a second reference sequence using the collected third historical monitoring data;

[0302] The standardization submodule is used to standardize both the second comparison sequence and the second reference sequence.

[0303] The calculation submodule is used to calculate the absolute difference between the standardized second comparison series and the standardized second reference series corresponding to each preset feature index;

[0304] The determination submodule is used to determine the gray-scale correlation between the standardized second comparison series and the standardized second reference series corresponding to each preset feature index using the absolute difference.

[0305] The selection submodule is used to select the feature indicators corresponding to the gray-scale correlation degree that meet the preset conditions as the second key early warning indicators;

[0306] The third historical monitoring data was collected earlier than the second historical monitoring data.

[0307] In some embodiments, the second construction submodule is specifically used for:

[0308] Data cleaning was performed on the third historical monitoring data;

[0309] Standardize the third historical monitoring data after cleaning the data corresponding to the preset feature indicators to obtain the standardized third historical monitoring data corresponding to the preset feature indicators.

[0310] Let the minimum value in the standardized third historical monitoring data corresponding to each preset feature indicator be the optimal value, and the maximum value in the standardized third historical monitoring data corresponding to each preset feature indicator be the worst value.

[0311] A second reference sequence is constructed using the optimal and worst values ​​corresponding to the preset feature indicators.

[0312] In some embodiments, a submodule is determined, specifically for:

[0313] Based on the absolute difference, determine the maximum and minimum absolute difference between the standardized second comparison series and the standardized second reference series corresponding to each preset feature index;

[0314] Using the maximum and minimum absolute differences, the gray-scale correlation degree between the standardized second comparison series and the standardized second reference series corresponding to each preset feature index is calculated.

[0315] In some embodiments, the fifth acquisition subunit includes:

[0316] The third construction submodule is used to extract data corresponding to the second key early warning indicator from the second historical monitoring data after data cleaning, and to construct the second feature set matrix.

[0317] The fourth construction submodule is used to construct the second new feature data based on the second feature set matrix using the principal component analysis method.

[0318] In some embodiments, the fourth construction submodule is specifically used for:

[0319] The second covariance matrix is ​​calculated using the second feature set matrix;

[0320] Calculate the cumulative contribution rate of the first p' features in the second covariance matrix;

[0321] When the cumulative contribution rate of the current p' features is greater than or equal to the expected value of the cumulative contribution rate, the second new feature data is constructed using the first p' features in the second covariance matrix;

[0322] Where p' is a positive integer, p'≤k', and k' is the total number of features in the second covariance matrix.

[0323] In some embodiments, determining the subunit includes:

[0324] The third acquisition module is used to extract valid data from the second new feature data based on the preset standard limit value of the transformer area during normal operation and the preset standard limit value of the transformer area during fault operation corresponding to the second key early warning indicator, and obtain the processed second new feature data.

[0325] The fourth acquisition module is used to perform feature clustering on the processed second new feature data based on the second key early warning indicator using the K-means clustering method, so as to obtain the clustered second new feature data.

[0326] The fifth acquisition module is used to analyze and obtain the extreme values ​​of change corresponding to each second key early warning indicator based on the second new feature data after clustering using the interval prediction method;

[0327] The prediction module is used to predict the sample variance of the Bootstrap sample corresponding to each second key early warning indicator based on the clustered second new feature data and the extreme values ​​of change corresponding to each second key early warning indicator using the Bootstrap algorithm.

[0328] The fourth calculation module is used to calculate the upper and lower limits of each second key early warning indicator by utilizing the sample variance corresponding to each second key early warning indicator.

[0329] The sixth acquisition module is used to determine the historical warning level of the distribution area corresponding to the second new feature data based on the upper and lower limits of each second key warning indicator.

[0330] In some embodiments, the sixth acquisition module is specifically used for:

[0331] When the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the first confidence threshold, the historical early warning level of the distribution area is the first early warning level.

[0332] When the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the second confidence threshold, the historical early warning level of the distribution area is the second early warning level.

[0333] When the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the third confidence threshold, the historical early warning level of the distribution area is the third early warning level.

[0334] When the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the fourth confidence threshold, the historical early warning level of the distribution area is the fourth early warning level.

[0335] In some embodiments, the sixth acquisition subunit is specifically used for:

[0336] The dataset is divided into a training set, a validation set, and a test set;

[0337] The long and short neural network model is trained using the training set to obtain the trained long and short neural network model.

[0338] The trained long and short neural network model is validated using a validation set. If the validation is successful, the performance of the trained long and short neural network model is evaluated using a test set. If the validation fails, the hyperparameters of the long and short neural network model are adjusted, overfitting is monitored, and the long and short neural network model is retrained until the validation is successful.

[0339] The performance of the trained long and short neural network model is evaluated using a test set. If the performance evaluation of the trained long and short neural network model is successful, then the trained long and short neural network model is an early warning model; if the performance evaluation of the trained long and short neural network model fails, then the trained long and short neural network model does not contain the features of that transformer area.

[0340] In some embodiments, the formula for calculating the first covariance matrix includes: B = X T X

[0341] The formula for calculating the cumulative contribution rate of the first p features in the first covariance matrix includes:

[0342] In the above formula, X is the first feature set matrix, T is the transpose, and B is the first covariance matrix; i∈[1,k], k is the total number of features in the first covariance matrix, p≤k; α P λ represents the cumulative contribution rate of the first p features in the first covariance matrix. i Let be the eigenvalue of the i-th feature in the first covariance matrix.

[0343] In some embodiments, the formula for calculating the second covariance matrix includes: B′=X′ T X′

[0344] The formula for calculating the cumulative contribution rate of the first p' features in the second covariance matrix includes:

[0345] In the above formula, X' is the second feature set matrix, T is the transpose, and B' is the second covariance matrix; i′∈[1,k′], k' is the total number of features in the second covariance matrix, p'≤k'; α′ P λ represents the cumulative contribution rate of the first p' features in the second covariance matrix. i ′ represents the eigenvalue of the i'th feature in the second covariance matrix.

[0346] In some embodiments, the formula for calculating the upper limit of each second key warning indicator includes:

[0347] The calculation formulas for the lower limits of each of the second key early warning indicators include:

[0348] In the above formula, Y up Y represents the upper limit of each of the second key early warning indicators. low These are the lower limits for each of the second key early warning indicators. Var represents the mean of the second new feature data after clustering for each of the second key early warning indicators. * (y) represents the sample variance corresponding to each of the second key early warning indicators, Z 1-α / 2 α is the critical value corresponding to 100·(1-α)% of the standard normal distribution, where α is the adjustment coefficient.

[0349] In some embodiments, the distribution radio area is a distribution radio area that includes electric vehicle charging stations.

[0350] It is understood that the system embodiments provided above correspond to the method embodiments described above, and the specific details can be referred to each other, which will not be repeated here.

[0351] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0352] As shown in Figure 3, this application also provides an electronic device 300, which may be a computer device, a microcontroller device, a smart mobile device, etc. The electronic device in this embodiment may include a processor 301, a memory 302, a transceiver component 303, etc. The memory 302, the processor 301, and the transceiver component 303 are connected via a bus; the memory 302 can be used to store executable programs, and an exemplary executable program may include instructions; the processor 301 is used to execute the instructions stored in the memory 302. The memory 302 can also be used to store data, which can be accessed and / or modified when instructions are executed.

[0353] The processor 301 may be a central processing unit (CPU), or it may 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. It is the computing core and control core of the terminal, and it is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions in the storage medium to realize the corresponding method flow or corresponding function, so as to realize the steps of the distribution area risk classification and early warning method in the above embodiment.

[0354] Based on the same inventive concept, this application also provides a readable storage medium, specifically an electronic device readable storage medium (Memory). This readable storage medium is a memory device within an electronic device, used to store programs and data. It is understood that the storage medium here can include both built-in storage media within the electronic device and extended storage media supported by the electronic device. The storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more executable programs (including program code). It should be noted that the storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. Loading and executing one or more instructions stored in the storage medium by the processor can implement the steps of the distribution area risk classification and early warning method in the above embodiments.

[0355] This application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program. When the computer program is read and executed by a computer, it implements some or all of the steps in the above-described method. This computer program product can be implemented specifically through hardware, software, or a combination thereof. In some embodiments, the computer program product is specifically embodied as a computer storage medium; in other embodiments, the computer program product is specifically embodied as a software product, such as a software development kit (SDK), etc.

[0356] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0357] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more flowchart illustrations and / or one or more block diagrams.

[0358] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0359] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0360] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and not to limit them. Although this application has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of this application. Any modifications or equivalent substitutions that do not depart from the spirit and scope of this application should be covered within the protection scope of the claims of this application. Industrial applicability

[0361] This application provides a method, system, device, storage medium, and program product for risk classification and early warning of distribution transformer substations. The method includes: extracting features from target monitoring data of the distribution transformer substation collected based on a first data acquisition time interval to obtain first substation data features; performing correlation analysis on preset feature indicators based on the first substation data features to obtain a first key early warning indicator; fusing feature indicators using the first key early warning indicator to obtain first new feature data; predicting the early warning level of the distribution transformer substation using an early warning model based on the first new feature data; adjusting the first data acquisition time interval according to the early warning level of the distribution transformer substation to obtain a second data acquisition time interval; re-acquiring target monitoring data based on the second data acquisition time interval, and re-obtaining the early warning level of the distribution transformer substation based on the re-acquired target monitoring data. This application enables flexible adjustment of the acquisition interval of the distribution transformer substation, improving the accuracy and reliability of risk early warning for distribution transformer substations.

Claims

1. A method for risk classification and early warning of distribution radio areas, comprising: Feature extraction is performed on the target monitoring data of the distribution station area to obtain the data features of the first distribution station area; The target monitoring data of the distribution radio area is collected based on the first data acquisition time interval; Based on the data characteristics of the first transformer area, a correlation analysis is performed on the preset feature indicators to obtain the first key early warning indicator. The first key early warning indicator is used to fuse feature indicators to obtain the first new feature data; Based on the first new feature data, the warning level of the distribution radio area is predicted using the warning model. Based on the warning level of the distribution radio area, the first data acquisition time interval is adjusted to obtain the second data acquisition time interval; The target monitoring data is re-acquired based on the second data acquisition time interval, and the warning level of the distribution radio station is re-determined based on the re-acquired target monitoring data.

2. The method according to claim 1, wherein, The step of extracting features from the target monitoring data of the distribution station area collected based on the first data acquisition time interval to obtain the data features of the first distribution station area includes: The target monitoring data is cleaned. Data corresponding to preset feature indicators are extracted from the cleaned target monitoring data to obtain the data features of the first transformer area.

3. The method according to claim 1 or 2, wherein, The first key early warning indicator is obtained by performing correlation analysis on preset feature indicators based on the data characteristics of the first transformer area, including: Based on the data characteristics of the first transformer area, the gray-scale correlation analysis method is used to perform correlation analysis on the preset feature indicators to obtain the first key early warning indicator.

4. The method according to claim 3, wherein, Based on the data characteristics of the first transformer area, a gray-scale correlation analysis method is used to perform correlation analysis on the preset feature indicators to obtain the first key early warning indicator, including: A first comparison sequence is constructed using the data features of the first transformer area; A first reference sequence is constructed using the first historical monitoring data collected. The first comparison sequence and the first reference sequence are standardized. For each preset feature index, determine the absolute difference between the first standardized comparison sequence corresponding to the preset feature index and the first standardized reference sequence corresponding to the feature index. For each preset feature index, the gray-level correlation degree between the first comparison sequence after standardization of the preset feature index and the first reference sequence after standardization of the feature index is determined by using the absolute difference corresponding to the feature index. The preset feature index corresponding to the gray-scale correlation degree that meets the preset conditions is selected as the first key early warning index.

5. The method according to claim 4, wherein, The construction of the first reference sequence using the collected first historical monitoring data includes: Perform data cleaning on the first historical monitoring data; Standardize the first historical monitoring data after data cleaning corresponding to the preset feature indicators to obtain the first historical monitoring data after standardization corresponding to the preset feature indicators. Let the minimum value in the first historical monitoring data after standardization corresponding to each preset feature index be the optimal value, and the maximum value in the first historical monitoring data after standardization corresponding to each preset feature index be the worst value. The first reference sequence is constructed using the optimal and worst values ​​corresponding to the preset feature indicators.

6. The method according to claim 4 or 5, wherein, The step of determining the gray-level correlation between the first standardized comparison series corresponding to the feature index and the first standardized reference series corresponding to the feature index by using the absolute difference corresponding to the feature index includes: Based on the absolute difference, determine the maximum absolute difference and minimum absolute difference between the first standardized comparison series corresponding to the preset feature index and the first standardized reference series corresponding to the preset feature index. Using the maximum and minimum absolute differences, the gray-scale correlation between the first comparison series and the first reference series after standardization, corresponding to the preset feature index, is determined.

7. The method according to claim 2, wherein, The step of fusing feature indicators using the first key early warning indicator to obtain the first new feature data includes: Extract data corresponding to the first key early warning indicator from the cleaned target monitoring data, and construct a first feature set matrix; Based on the first feature set matrix, the first new feature data is constructed using principal component analysis.

8. The method according to claim 7, wherein, The construction of the first new feature data based on the first feature set matrix using principal component analysis includes: The first covariance matrix is ​​determined using the first feature set matrix; Determine the cumulative contribution rate of the first p features in the first covariance matrix; When the cumulative contribution rate of the first p features is greater than or equal to the expected value of the cumulative contribution rate, the first new feature data is constructed using the first p features in the first covariance matrix. Where p is a positive integer, p≤k, and k is the total number of features in the first covariance matrix.

9. The method according to claim 2, wherein, The step of adjusting the first data acquisition time interval according to the warning level of the distribution radio area to obtain the second data acquisition time interval includes: When the warning level is the fourth warning level or the number of electric vehicles entering the charging station in the distribution area reaches the first percentage of the charging station capacity, the second data collection time interval is the first multiple of the first data collection time interval. When the warning level is the third warning level or the number of electric vehicles entering the charging station in the distribution area reaches the second percentage of the charging station capacity, the second data collection time interval is the second multiple of the first data collection time interval. When the warning level is lower than or equal to the second warning level, and the number of electric vehicles entering the distribution area is less than the second percentage of the charging station capacity, the second data collection time interval is a third multiple of the first data collection time interval. When the warning level is lower than or equal to the first warning level, and the number of electric vehicles entering the distribution area is less than the second percentage of the charging station capacity, the second data collection time interval is the first data collection time interval.

10. The method according to any one of claims 1 to 9, wherein, The process of establishing the early warning model includes: Collect the second historical monitoring data of the distribution radio station area; Feature extraction is performed on the second historical monitoring data to obtain the data features of the second transformer area; Based on the data characteristics of the second transformer area, a correlation analysis is performed on the preset feature indicators to obtain the second key early warning indicator. The second key early warning indicator is used to fuse feature indicators to obtain a second new feature data; Determine the historical early warning level of the distribution radio station area corresponding to the second new feature data; A dataset is constructed using the second new feature data and the historical warning levels of the distribution radio area, and the long and short neural network model is trained using the dataset to obtain the warning model.

11. The method according to claim 10, wherein, The feature extraction of the second historical monitoring data to obtain the features of the second transformer area data includes: Data cleaning is performed on the second historical monitoring data; Data corresponding to preset feature indicators are extracted from the second historical monitoring data after data cleaning to obtain the data features of the second transformer area.

12. The method according to claim 10, wherein, The second key early warning indicator is obtained by performing correlation analysis on preset feature indicators based on the data characteristics of the second transformer area, including: Based on the data characteristics of the second transformer area, the gray-scale correlation analysis method is used to perform correlation analysis on the preset feature indicators to obtain the second key early warning indicator.

13. The method according to claim 12, wherein, Based on the data characteristics of the second transformer area, a gray-scale correlation analysis method is used to perform correlation analysis on preset feature indicators to obtain the second key early warning indicator, including: A second comparison sequence is constructed using the data features of the second transformer area; A second reference series was constructed using the collected third historical monitoring data; Both the second comparison sequence and the second reference sequence are standardized. For each preset feature index, determine the absolute difference between the standardized second comparison sequence corresponding to the feature index and the standardized second reference sequence corresponding to the feature index. For each preset feature index, the gray-level correlation degree between the standardized second comparison sequence corresponding to the preset feature index and the standardized second reference sequence corresponding to the feature index is determined by using the absolute difference corresponding to the feature index. Select the feature index corresponding to the gray-scale correlation degree that meets the preset conditions as the second key early warning index; The third historical monitoring data was collected earlier than the second historical monitoring data.

14. The method according to claim 13, wherein, The construction of the second reference series using the collected third historical monitoring data includes: The third historical monitoring data is cleaned. The third historical monitoring data after data cleaning corresponding to the preset feature index is standardized to obtain the standardized third historical monitoring data corresponding to the preset feature index. Let the minimum value in the standardized third historical monitoring data corresponding to each preset feature indicator be the optimal value, and the maximum value in the standardized third historical monitoring data corresponding to each preset feature indicator be the worst value. The second reference sequence is constructed using the optimal and worst values ​​corresponding to the preset feature indicators.

15. The method according to claim 13 or 14, wherein, The step of determining the grayscale correlation between the standardized second comparison series corresponding to the preset feature index and the standardized second reference series corresponding to the feature index by using the absolute difference corresponding to the feature index includes: Based on the absolute difference, determine the maximum absolute difference and minimum absolute difference between the standardized second comparison series corresponding to the preset feature index and the standardized second reference series corresponding to the preset feature index; Using the maximum and minimum absolute differences, the gray-scale correlation between the standardized second comparison series and the standardized second reference series corresponding to the preset feature index is determined.

16. The method according to claim 11, wherein, The process of fusing feature indicators using the second key early warning indicator to obtain second new feature data includes: Extract data corresponding to the second key early warning indicator from the second historical monitoring data after data cleaning, and construct a second feature set matrix; Based on the second feature set matrix, the second new feature data is constructed using principal component analysis.

17. The method according to claim 16, wherein, The construction of the second new feature data based on the second feature set matrix using principal component analysis includes: The second covariance matrix is ​​determined using the second feature set matrix; Determine the cumulative contribution rate of the first p' features in the second covariance matrix; When the cumulative contribution rate of the current p' features is greater than or equal to the expected value of the cumulative contribution rate, the second new feature data is constructed using the first p' features in the second covariance matrix; Where p' is a positive integer, p'≤k', and k' is the total number of features in the second covariance matrix.

18. The method according to any one of claims 10 to 16, wherein, Determining the historical early warning level of the distribution radio area corresponding to the second new feature data includes: Based on the preset standard limit values ​​for normal operation of the transformer area corresponding to the second key early warning indicator, and the preset standard limit values ​​for fault operation of the transformer area, the effective data in the second new feature data is extracted to obtain the processed second new feature data. Based on the second key early warning indicator, the K-means clustering method is used to perform feature clustering on the processed second new feature data to obtain the clustered second new feature data. Based on the second new feature data after clustering, the extreme values ​​of change corresponding to each second key early warning indicator are obtained by interval prediction method. Based on the clustered second new feature data and the extreme values ​​of change corresponding to each second key early warning indicator, the sample variance of the Bootstrap sample corresponding to each second key early warning indicator is predicted using the Bootstrap algorithm. Using the sample variance corresponding to each of the second key early warning indicators, the upper and lower limits of each second key early warning indicator are determined; Based on the upper and lower limits of each of the second key early warning indicators, the historical early warning level of the distribution area corresponding to the second new feature data is determined.

19. The method according to claim 18, wherein, The step of determining the historical early warning level of the distribution area corresponding to the second new feature data based on the upper and lower limits of each of the second key early warning indicators includes: If the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the first confidence threshold, the historical early warning level of the distribution area is the first early warning level. If the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the second confidence threshold, the historical early warning level of the distribution area is the second early warning level. If the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the third confidence threshold, the historical early warning level of the distribution area is the third early warning level. If the upper and lower limits of the second key early warning indicator fall within the fluctuation range corresponding to the fourth confidence threshold, the historical early warning level of the distribution area is the fourth early warning level.

20. The method according to any one of claims 10 to 19, wherein, The step of training the long and short neural network model using the dataset to obtain the early warning model includes: The dataset is divided into a training set, a validation set, and a test set; The long and short neural network model is trained using the training set to obtain the trained long and short neural network model; The trained long and short neural network model is validated using a validation set. If the validation is successful, the performance of the trained long and short neural network model is evaluated using the test set. If the validation fails, the hyperparameters of the long and short neural network model are adjusted, overfitting is monitored, and the long and short neural network model is retrained until the validation is successful. The performance of the trained long and short neural network model is evaluated using the test set. If the performance evaluation of the trained long and short neural network model is successful, then the trained long and short neural network model is the early warning model; if the performance evaluation of the trained long and short neural network model fails, then the trained long and short neural network model does not contain the features of that transformer area.

21. The method according to claim 8, wherein, The formula for calculating the first covariance matrix includes: B = X T X The formula for calculating the cumulative contribution rate of the first p features in the first covariance matrix includes: In the above formula, X is the first feature set matrix, T is the transpose, and B is the first covariance matrix; i∈[1,k], k is the total number of features in the first covariance matrix, p≤k; α P λ represents the cumulative contribution rate of the first p features in the first covariance matrix. i Let be the eigenvalue of the i-th feature in the first covariance matrix.

22. The method according to claim 17, wherein, The formula for calculating the second covariance matrix includes: B′=X′ T X′ The formula for calculating the cumulative contribution rate of the first p' features in the second covariance matrix includes: In the above formula, X' is the second feature set matrix, T is the transpose, and B' is the second covariance matrix; i′∈[1,k′], k' is the total number of features in the second covariance matrix, p'≤k'; α′ P λ′ represents the cumulative contribution rate of the first p' features in the second covariance matrix. i Let be the eigenvalue of the i'th feature in the second covariance matrix.

23. The method according to claim 18, wherein, The calculation formulas for the upper limits of each of the second key early warning indicators include: The calculation formulas for the lower limits of each of the second key early warning indicators include: In the above formula, Y up Y represents the upper limit of each of the second key early warning indicators. low These are the lower limits for each of the second key early warning indicators. Var is the mean of the second new feature data after clustering corresponding to each second key early warning indicator. * (y) represents the sample variance corresponding to each of the second key early warning indicators, Z 1-α / 2 α is the critical value corresponding to 100·(1-α)% of the standard normal distribution, where α is the adjustment coefficient.

24. The method according to claim 1, wherein, The distribution radio area refers to a distribution radio area that includes electric vehicle charging stations.

25. A risk classification and early warning system for power distribution areas, comprising: The extraction unit is used to extract features from the target monitoring data of the distribution station area to obtain the first area data features; The target monitoring data of the distribution radio area is collected based on the first data acquisition time interval; The first acquisition unit is used to perform correlation analysis on preset feature indicators based on the data characteristics of the first transformer area to obtain the first key early warning indicator. The second acquisition unit is used to fuse feature indicators using the first key early warning indicator to obtain the first new feature data. The third acquisition unit is used to predict the warning level of the distribution radio station area based on the first new feature data and using the warning model. The adjustment unit is used to adjust the first data acquisition time interval according to the warning level of the distribution radio area to obtain the second data acquisition time interval; The fourth acquisition unit is used to reacquire the target monitoring data based on the second data acquisition time interval, and to redetermine the warning level of the distribution radio station area based on the reacquired target monitoring data.

26. The system according to claim 25, wherein, The extraction unit includes: The data cleaning subunit is used to clean the target monitoring data. The first acquisition subunit is used to extract data corresponding to preset feature indicators from the cleaned target monitoring data to obtain the data features of the first transformer area.

27. The system according to claim 25, wherein, The first acquisition unit includes: The second acquisition subunit is used to perform correlation analysis on the preset feature indicators based on the data characteristics of the first transformer area using a grayscale correlation analysis method to obtain the first key early warning indicator.

28. The system according to claim 27, wherein, The second acquisition subunit includes: The first construction module is used to construct a first comparison sequence using the data features of the first transformer area; The second construction module is used to construct a first reference sequence using the collected first historical monitoring data; The processing module is used to standardize both the first comparison sequence and the first reference sequence; The first calculation module is used to determine the absolute difference between the first comparison sequence after standardization of the preset feature index and the first reference sequence after standardization of the feature index for each preset feature index. The determination module is used to determine the gray-level correlation between the first comparison sequence after standardization of the preset feature index and the first reference sequence after standardization of the feature index for each preset feature index by using the absolute difference corresponding to the feature index. The selection module is used to select a preset feature index corresponding to the grayscale correlation degree that meets preset conditions as the first key early warning index.

29. The system according to claim 28, wherein, The second building module is specifically used for: Perform data cleaning on the first historical monitoring data; Standardize the first historical monitoring data after data cleaning corresponding to the preset feature indicators to obtain the first historical monitoring data after standardization corresponding to the preset feature indicators. Let the minimum value in the first historical monitoring data after standardization corresponding to each preset feature index be the optimal value, and the maximum value in the first historical monitoring data after standardization corresponding to each preset feature index be the worst value. The first reference sequence is constructed using the optimal and worst values ​​corresponding to the preset feature indicators.

30. The system according to claim 28, wherein, The determining module is specifically used for: Based on the absolute difference, determine the maximum absolute difference and minimum absolute difference between the first standardized comparison series corresponding to the preset feature index and the first standardized reference series corresponding to the preset feature index. Using the maximum and minimum absolute differences, the gray-scale correlation between the first comparison series and the first reference series after standardization, corresponding to the preset feature index, is determined.

31. The system according to claim 26, wherein, The second acquisition unit includes: The first construction subunit is used to extract data corresponding to the first key early warning indicator from the target monitoring data after data cleaning, and construct the first feature set matrix. The second construction subunit is used to construct the first new feature data based on the first feature set matrix using principal component analysis.

32. The system according to claim 31, wherein the second building subunit comprises: The second calculation module is used to determine the first covariance matrix using the first feature set matrix; The third calculation module is used to determine the cumulative contribution rate of the first p features in the first covariance matrix; The third construction module is used to construct the first new feature data using the first p features in the first covariance matrix when the cumulative contribution rate of the first p features is greater than or equal to the expected value of the cumulative contribution rate. Where p is a positive integer, p≤k, and k is the total number of features in the first covariance matrix.

33. The system according to claim 25, wherein, The adjustment unit is specifically used for: When the warning level is the fourth warning level or the number of electric vehicles entering the charging station in the distribution area reaches the first percentage of the charging station capacity, the second data collection time interval is the first multiple of the first data collection time interval. When the warning level is the third warning level or the number of electric vehicles entering the charging station in the distribution area reaches the second percentage of the charging station capacity, the second data collection time interval is the second multiple of the first data collection time interval. When the warning level is lower than or equal to the second warning level, and the number of electric vehicles entering the distribution area is less than the second percentage of the charging station capacity, the second data collection time interval is a third multiple of the first data collection time interval. When the warning level is lower than or equal to the first warning level, and the number of electric vehicles entering the distribution area is less than the second percentage of the charging station capacity, the second data collection time interval is the first data collection time interval.

34. The system according to claim 25, wherein, The distribution radio area refers to a distribution radio area that includes electric vehicle charging stations.

35. An electronic device comprising: At least one processor and memory; The memory and processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the method for classifying and warning of transformer area risks as described in any one of claims 1 to 24 is implemented.

36. A readable storage medium having an executable program stored thereon, wherein when the executable program is executed, it implements the transformer area risk classification and early warning method as described in any one of claims 1 to 24.

37. A computer program product comprising a computer program or instructions, wherein when the computer program or instructions are executed on a computer, the computer executes the method for classifying and warning of transformer area risks as described in any one of claims 1 to 24.

Images