Method, device and equipment for evaluating out-of-tolerance risk of multiple charging piles at same station and medium

Abstract

The invention discloses a same-station multi-charging-pile out-of-tolerance risk assessment method, device and equipment and a medium, and the method comprises the steps: S1, data collection and preprocessing, S2, charging pile metering error change curve segmentation and high-dimensional feature matrix construction: segmenting the error change curve of each charging pile according to a periodic proportion, and constructing a high-dimensional feature matrix; constructing a high-dimensional feature matrix corresponding to each segment; s3, principal component analysis dimensionality reduction; s4, performing clustering and out-of-tolerance risk assessment based on low-dimensional features, and dividing a normal charging pile, a to-be-observed charging pile and a high-risk charging pile according to a cluster proportion and a distance threshold; and S5, risk assessment result verification: in combination with actual verification data at the end of the verification cycle, verifying the reliability of the risk assessment result by calculating the high-risk identification accuracy and the normal charging pile missed judgment rate. The problems that in the prior art, out-of-tolerance risk identification lags behind, the cost is high due to the fact that a pre-calibration standard charging pile or large-scale training data is relied on, manual intervention of the evaluation process is large, and control pertinence is insufficient are solved.

Description

Technical Field

[0001] This application relates to the field of electric vehicle charging pile metering and testing technology, and in particular, to a method, device, equipment and medium for assessing the risk of multiple charging piles exceeding tolerances at the same station. Background Technology

[0002] With the rapid development of the electric vehicle (EV) industry, public charging piles, as a core infrastructure for energy replenishment, are directly related to the fairness of charging transactions, the protection of user rights, and the compliant operation of operators due to their metering accuracy. Industry data shows that global electric vehicle sales reached 17.1 million units in 2024, accounting for 21% of total global vehicle sales; as of January 2025, the number of public charging piles in China had reached 3.76 million, including 1.71 million DC charging piles and 2.04 million AC charging piles. Similar to fuel dispensers, charging piles, as measuring instruments used for trade settlement, must comply with the OIML G22 standard issued by the International Organization of Legal Metrology (OIML) and the mandatory verification procedures of China, such as JJG1148 (AC charging piles) and JJG1149 (DC charging piles), to ensure metering accuracy.

[0003] Currently, the performance evaluation of charging pile metering mainly relies on traditional on-site verification methods. This requires professionals to bring standard charging piles to the site for operation, which is not only inefficient and costly, but also has limited coverage. Evaluations can only be conducted at the end of a fixed verification cycle, leading to delays in identifying out-of-tolerance risks and potentially causing unfair transactions and compliance risks. Furthermore, the evaluation process involves significant manual intervention and lacks targeted control, making it difficult to meet the needs of dynamic monitoring and early warning for large-scale charging pile networks. In addition, although existing research has explored charging pile metering error analysis and anomaly early warning, resulting in some evaluation methods, these methods rely on pre-calibrated standard charging piles or large-scale, high-quality training data, leading to high implementation costs and a difficulty in balancing evaluation accuracy and efficiency. Summary of the Invention

[0004] This application provides a method for assessing the risk of out-of-tolerance issues among multiple charging piles at the same station, which solves the technical problems of existing charging pile metering performance assessment processes, such as excessive manual intervention and insufficient targeted control leading to slow and inefficient identification of out-of-tolerance risks, and high costs due to reliance on pre-calibrated standard charging piles or large-scale training data.

[0005] This application is achieved through the following solution: A method for assessing the risk of multiple charging piles exceeding tolerances at the same station includes the following steps: S1. Collect measured current and voltage data of multiple charging piles at the same station within a single calibration cycle, obtain calibration standard values ​​through high-precision measurement, calculate the instantaneous error and daily average error, and construct the error change curve of a single charging pile; S2. Divide the error change curve of each charging pile into segments according to the period ratio, eliminate the difference in the initial error magnitude of different charging piles, and construct the high-dimensional feature matrix corresponding to each segment. S3. The high-dimensional feature matrix of each error change curve is reduced by PCA. The principal components and their corresponding variance contributions are obtained by covariance matrix calculation and eigenvalue decomposition. Core features are extracted to construct a low-dimensional feature matrix. S4. Execute the K-Means unsupervised clustering algorithm based on the low-dimensional feature matrix to cluster all charging piles within the station. p Charging piles with similar segment measurement error characteristic curves and error change curves are grouped into the same cluster. The optimal number of clusters and the core trend clusters are determined. Based on the cluster ratio and distance threshold, charging piles are divided into three categories: normal, unobservable, and high-risk. S5. At the end of the verification cycle, based on clock synchronization, the verification standard power value is compared with the actual measured power value of the charging pile to obtain the actual verification data of each charging pile. The reliability of the risk assessment results is verified by calculating the high-risk identification accuracy rate and the normal charging pile false detection rate, thus forming a closed-loop management system.

[0006] Further, step S1 specifically includes the following steps: S11. Raw metering data collection: Collect the first data from the same station daily at a preset frequency. Measured current value of pile cap Measured voltage value , t Timestamps within each day , This refers to the number of times data is collected per day. S12. Collection of verification standard values: On the first day of the verification cycle ( d =1), determined by a high-level measurement method, the first Current verification standard value of charging piles in Taiwan Voltage verification standard value (This standard value is fixed within the cycle and does not require daily calibration, making it suitable for verification cycles of different lengths.) S13. Intraday Error Calculation: Calculate the following formula: i Taiwan charging pile d The first day t Instantaneous measurement error: ; In the formula, For the first i Taiwan charging pile d The first day t Sub-instantaneous measurement error; S14. Calculation of daily average error: For the first... d The sky The daily average error is obtained by averaging the instantaneous errors. To avoid interference from intraday high-frequency fluctuations in trend analysis: ; S15. Definition of error variation curve: based on the number of days within a single verification cycle. d The horizontal axis represents the average daily error. Using the vertical axis, construct the first i Error variation curve of a charging pile within a single calibration cycle : ; Among them, the error change curve For discrete points, reflecting the first point within the station i The trend of error over time in a single calibration cycle of the charging pile.

[0007] Furthermore, step S2 specifically includes the following steps: S21. Divide the error variation curve of each charging pile into periods proportionally. m The error variation curve of ≥2 consecutive segments is defined as the first segment. p The cumulative number of days corresponding to the segment error change curve is : ; The above constraints are: , D The total number of days in the calibration cycle for a single charging pile within the same station; When the cumulative number of days At that time, the integration of the first i In front of the charging station Error sequence for days: ; Within the same site i Taiwan charging pile p The error variation curves within each segmented interval are as follows: ; S22. For the error variation curve of each charging pile in the above station, normalization is performed using the following formula to eliminate the impact of differences in the initial error magnitude of different charging piles on the trend comparison: ; In the formula: , They are respectively Minimum and maximum values; based on N Taiwan charging pile p Segment error change curve Construct the high-dimensional feature matrix corresponding to this segment. : ; in, Indicates the first The number of days included in the segment error variation curve, i.e., the feature dimension; each element in the matrix. Indicates the first The charging station, on the first The first segment error change curve Error value after normalization .

[0008] Furthermore, step S3 specifically includes the following steps: S31. Standardization of Error Variation Curve Characteristics: For high-dimensional matrices Each column is Z-score standardized to eliminate the interference of variance differences in features across different days on the dimensionality reduction results, ensuring fair contributions of each feature to the principal components and adapting to feature distributions across different testing periods. The formula is as follows: ; In the formula: For the first d The mean of the column features; For the first d Standard deviation of the column features; S32. Covariance Matrix and Eigenvalue Decomposition: List the... p Covariance matrix corresponding to segment error variation curve data : ; The formula for calculating each element is as follows: ; In the formula: Indicates the site N The first charging pile p The first segment of the measurement error change curve a Heaven and the First b Covariance of the error characteristics; For covariance matrix Eigenvalue decomposition yields eigenvalues ​​and eigenvectors: ; Among them, the feature matrix Elements in the matrix Representing the p The first segment of the measurement error change curve The principal component in the first... dWeights on the eigenvalue feature; eigenvalue diagonal matrix , It is the covariance matrix eigenvalues, For the first s The eigenvalues ​​corresponding to the principal components, and satisfying ; S33, Principal Component Selection: From D p Before selecting from the principal components k (p) One makes the front k (p) The sum of the variance contributions of each principal component accounts for no less than the proportion of the total contribution of all principal components. , To retain a threshold for information, k (p) = D p , The formula is: ; Combination The above formula can be simplified to: ; S34. Construction of Low-Dimensional Feature Matrix: Determining k (p) Then, from the feature matrix U (p) Extracting the first k (p) column vectors constitute Dimensionality reduction matrix : ; In the formula: elements of the matrix Indicates the first p The first segment error change curve j The principal components after dimensionality reduction are in the... d The weights of the error characteristics of each day are obtained by linear combination. j Low-dimensional features; Standardized high-dimensional matrix With dimensionality reduction matrix Multiply, we get low-dimensional feature matrix : ; low-dimensional feature matrix It can be expanded as follows: ; In the formula, For the first time on the site i Taiwan in p The low-dimensional eigenvector of the segment measurement error variation curve; Indicates the first i The number of charging piles in the first p The first segment of the measurement error change curve j The low-dimensional features are calculated by the following linear combination: .

[0009] Further, in step S33, from D p Before selecting from the principal components k (p) When there are multiple options, choose the smallest one. k (p) ,1≤ k (p) ≤ D p Make the former k (p) The sum of the variance contributions of each principal component accounts for no less than the proportion of the total contribution of all principal components. θ When θ is taken to be .

[0010] Furthermore, step S4 specifically includes the following steps: S41. Using the K-Means unsupervised clustering algorithm, all charging piles in the station are clustered. p Charging piles with similar segment measurement error characteristic curves and error change curves are grouped into the same cluster: Optimal number of clusters Determine: The range of the number of clusters C is... , , This is suitable for a typical scale of N charging piles; the optimal value is selected through the sum of squared errors within clusters (SSE), and the SSE formula is: ; In the formula, For the first c The set of charging station indexes for a cluster. For the first c Cluster center ; The smaller the value, the more consistent the low-dimensional characteristics of the charging piles within the cluster; Execute the K-Means unsupervised clustering algorithm: for low-dimensional feature matrices To perform clustering and avoid local optima caused by random initialization, the steps are as follows: Initialize cluster centers: Select using the K-Means algorithm CInitial center Ensure that the initial cluster centers are distributed to cover the different error variation curves of the characteristic curve of the p-th segment of the charging piles in the station; Sample allocation: For each low-dimensional feature vector Calculate its relationship with each cluster center Weighted Euclidean distance: ; In the formula, Let be the weight of the j-th feature, and t be the iteration number. Assign to the cluster with the smallest distance ; For each cluster The update center is the mean of the low-dimensional feature vectors of the charging piles within the cluster: ; In the formula, Let be the number of charging piles in the c-th cluster at the t-th iteration; Convergence criterion: Repeated sample assignment and center update until the following condition is met: or reaching the maximum number of iterations. , The convergence threshold is usually set to 1. ; S42. Determination of Out-of-Tolerance Risk Level: Based on the group consistency of the error variation curves of charging piles at the same station during the calibration cycle, and combined with clustering results, the out-of-tolerance risk of each charging pile is determined within the cycle. The risk level classification reflects general coverage of all charging piles, specifically including: Core trend cluster identification: p The sample proportion of the c-th cluster of the segment measurement error variation curve is: ; In the formula, the cluster with the largest proportion is the core trend cluster. ,Right now This represents the changing trend of the mainstream metering error curve of the charging piles at the same station; Set up risk assessment equations, including: Determining a Normal Charging Pile: A normal charging pile must simultaneously meet the following criteria: belonging to the core trend cluster and being relatively close to the center of the core trend cluster. This ensures that the error variation curve is highly consistent with that of most charging piles. The equation set is as follows: ; In the formula, As the core trend cluster center, Normal trend distance threshold , which is the criterion for Euclidean distance; High-risk charging pile judgment equations: A high-risk charging pile meets either the condition that its cluster size is too small or its distance from the center of the core trend cluster is too far, indicating that the error change curve deviates significantly from the group and there is a risk of exceeding the tolerance. The judgment equations are as follows: ; In the formula, The threshold for the proportion of small clusters is 0.1 ≤ r th ≤0.2, The high-risk trend distance threshold is used as the high-risk boundary of the Euclidean distance, where 1.0 ≤ d th2 ≤1.5, and d th2 > d th1 ; The equations for judging the charging piles to be observed are as follows: The charging pile to be observed must simultaneously meet the following conditions: it does not belong to the core trend cluster, its cluster is not a small cluster, its distance from the center of the core trend cluster is between two thresholds, and it is in a transitional state between normal and high risk. The equations are as follows: ; Based on the risk assessment equation, the final output is the risk level label for each charging pile at the same station, including normal charging piles, observation charging piles, and high-risk charging piles.

[0011] Furthermore, in step S5, calculating the high-risk identification accuracy rate specifically involves: ; In the formula, The number of charging piles that were predicted to be high-risk and whose deviations were confirmed at the end of the period; This refers to the total number of charging piles identified as high-risk throughout the entire lifecycle of the same station. Acc The target value is: Acc nomal ≥90%; The specific calculation of the missed detection rate of normal charging piles is as follows: ; In the formula: The number of charging piles that were initially considered normal but were confirmed to be out of tolerance at the end of the cycle; The total number of charging piles predicted to be normal throughout the entire cycle; target value .

[0012] This application also provides a device for assessing the risk of multiple charging piles exceeding tolerances at the same station, including: The data acquisition and preprocessing module is used to collect measured current and voltage data of multiple charging piles at the same station within a single calibration cycle. It obtains calibration standard values ​​(including voltage, current, power, and energy values) through high-precision measurement, calculates intraday instantaneous errors and daily average errors, and constructs an error variation curve for each charging pile. (At the end of the calibration cycle, the data acquisition module is also responsible for collecting measured power and energy data of the charging piles.) Calibration standard values ​​can also be obtained using external energy metering standard equipment. The segmentation and high-dimensional feature matrix construction module divides the error change curve of each charging pile into segments according to the period ratio, eliminates the difference in the initial error magnitude of different charging piles, and constructs the high-dimensional feature matrix corresponding to each segment. The principal component analysis dimensionality reduction module is used to perform PCA dimensionality reduction on high-dimensional feature matrices. It obtains each principal component and its corresponding variance contribution through covariance matrix calculation and eigenvalue decomposition, and extracts core features to construct a low-dimensional feature matrix. The low-dimensional feature-based clustering and out-of-range risk assessment module is used to execute the K-Means unsupervised clustering algorithm based on a low-dimensional feature matrix, and to cluster all charging piles in the station. p Charging piles with similar segment measurement error characteristic curves and error change curves are grouped into the same cluster. The optimal number of clusters and the core trend clusters are determined. Based on the cluster ratio and distance threshold, charging piles are divided into three categories: normal, unobservable, and high-risk. The risk assessment result verification module, at the end of the verification cycle, compares the standard electrical energy value with the actual measured electrical energy value of the charging pile based on clock synchronization to obtain the actual verification data for each charging pile. The reliability of the risk assessment results is verified by calculating the high-risk identification accuracy rate and the normal charging pile false negative rate, thus forming a closed-loop management system. (Actual verification data can also be obtained through external electrical energy metering standard equipment.) This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the method for assessing the risk of multiple charging piles exceeding tolerance at the same station.

[0013] This application also provides a storage medium including a stored program that, when the program is executed, controls the device where the storage medium is located to perform the steps of the method for assessing the risk of multiple charging piles exceeding tolerance at the same station.

[0014] Compared with the prior art, this application has the following advantages: 1. Addressing the issue that existing technologies can only identify out-of-tolerance charging piles at the end of the verification cycle, resulting in delayed risk detection, this application, based on the core characteristic of consistent operating conditions for multiple charging piles at the same station, uses full-cycle segmented monitoring and correlation analysis of error change curves of multiple piles to move the time node for identifying out-of-tolerance risks forward to within the verification cycle, breaking through the time limitations of the traditional model; at the same time, by using the error change curve deviation judgment mechanism, abnormal charging piles can be detected in advance, fundamentally solving the problems of unfair transactions, damage to user rights, and operator compliance risks caused by out-of-tolerance risks, and realizing the transformation from passive post-event handling to proactive pre-event warning; 2. To address the issue that existing technologies rely on large amounts of high-quality training data or additional standard charging piles, resulting in high deployment costs, this application utilizes only the measured current and voltage data from the daily operation of charging piles. Through a combination of segmented processing, PCA dimensionality reduction, and K-Means clustering algorithms, an error variation curve analysis model is constructed, eliminating the need for pre-calibration of standard charging piles, standard vehicles, and large-scale training data. Simultaneously, by extracting features and utilizing redundant data, it reduces upfront investment and data dependence while ensuring high-risk identification accuracy and low false negative rate. This overcomes the bottleneck of traditional evaluation methods where cost and accuracy are difficult to balance, and adapts to the efficient evaluation needs of large-scale charging pile networks. 3. This application provides a reliable method for assessing out-of-tolerance risks. It is an online assessment method for charging pile metering out-of-tolerance risks based on the correlation of error change curves among multiple charging piles at the same station. This method, through a closed-loop architecture design, achieves full automation of the entire process, including automatic raw data collection, intelligent calculation of daily average error, feature dimensionality reduction and clustering, automatic risk level determination, and quantitative verification of results. The core lies in preserving the stage-specific characteristics of the error change curve through segmented processing, combining PCA dimensionality reduction and K-Means clustering to extract the core trends of multiple charging piles at the same station, and innovatively introducing a three-level classification rule based on trend deviation (normal charging piles, observation charging piles, and high-risk charging piles), forming a continuous monitoring mechanism within a cycle. This method ultimately achieves early identification and precise control of charging pile out-of-tolerance risks under all operating conditions, significantly improving metering management efficiency, avoiding unfair transactions, and ensuring the accuracy of charging settlement and market fairness.

[0015] In addition to the purposes, features, and advantages described above, this application has other purposes, features, and advantages. A further detailed description of this application will be provided below with reference to the figures. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] 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, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein: Figure 1 This is a flowchart illustrating the method for assessing the risk of multiple charging piles exceeding tolerances at the same station, according to a preferred embodiment of this application. Figure 2 This is a schematic diagram of the metering error variation curves of 25 charging piles at the same site in a preferred embodiment of this application; Figure 3 This is the principal component clustering result of 25 charging piles within the same station in the preferred embodiment of this application; Figure 4 This is a schematic diagram of the multi-charging pile out-of-tolerance risk assessment device module of a preferred embodiment of this application; Figure 5 This is a schematic block diagram of an electronic device according to a preferred embodiment of this application; Figure 6 This is an internal structural diagram of a computer device according to a preferred embodiment of this application. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0019] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0020] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a device for assessing the risk of exceeding tolerances for multiple charging piles at the same station that can achieve the above functions. The following description uses a device for assessing the risk of exceeding tolerances for multiple charging piles at the same station as the executing entity to illustrate this embodiment and the following embodiments.

[0021] The following example uses a public charging station in the core area of ​​a city with 25 DC charging piles (N=25) deployed. The quarterly verification cycle is D=90, and the metering data is collected 6 times a day (f=6). It is necessary to identify out-of-tolerance risk equipment in advance within the verification cycle to avoid unfair transactions during the cycle.

[0022] like Figure 1 As shown, a preferred embodiment of this application provides a method for assessing the risk of multiple charging piles exceeding tolerances at the same station, including the following steps: S1. Collect measured current and voltage data of multiple charging piles at the same station within a single calibration cycle, obtain calibration standard values ​​through high-precision measurement, calculate the instantaneous error and daily average error, and construct the error change curve of a single charging pile; S2. Divide the error change curve of each charging pile into segments according to the period ratio, eliminate the difference in the initial error magnitude of different charging piles, and construct the high-dimensional feature matrix corresponding to each segment. S3. The high-dimensional feature matrix of each error change curve is reduced by PCA. The principal components and their corresponding variance contributions are obtained by covariance matrix calculation and eigenvalue decomposition. Core features are extracted to construct a low-dimensional feature matrix. S4. Execute the K-Means unsupervised clustering algorithm based on the low-dimensional feature matrix to cluster all charging piles within the station. p Charging piles with similar segment measurement error characteristic curves and error change curves are grouped into the same cluster. The optimal number of clusters and the core trend clusters are determined. Based on the cluster ratio and distance threshold, charging piles are divided into three categories: normal, unobservable, and high-risk. S5. At the end of the verification cycle, based on clock synchronization, the verification standard power value is compared with the actual measured power value of the charging pile to obtain the actual verification data of each charging pile. The reliability of the risk assessment results is verified by calculating the high-risk identification accuracy rate and the normal charging pile false detection rate, thus forming a closed-loop management system.

[0023] Compared with the prior art, this embodiment has the following beneficial effects: 1. Addressing the issue that existing technologies can only identify out-of-tolerance charging piles at the end of the calibration cycle, resulting in delayed risk detection, this embodiment leverages the core characteristic of consistent operating conditions across multiple charging piles at the same station. Through full-cycle segmented monitoring and correlation analysis of error change curves across multiple piles, it shifts the time node for identifying out-of-tolerance risks to within the calibration cycle, breaking through the time limitations of traditional methods. Simultaneously, by utilizing the deviation judgment mechanism of error change curves, it can detect abnormal charging piles in advance, fundamentally resolving transaction unfairness, user rights infringement, and operator compliance risks caused by out-of-tolerance hazards, thus achieving a shift from passive post-event handling to proactive pre-event warning. 2. To address the issue that existing technologies rely on large amounts of high-quality training data or additional standard charging piles, resulting in high deployment costs, this embodiment utilizes only the measured current and voltage data from the daily operation of charging piles. Through a combination of segmented processing, PCA dimensionality reduction, and K-Means clustering algorithms, an error variation curve analysis model is constructed, eliminating the need for pre-calibration of standard charging piles, standard vehicles, and large-scale training data. Simultaneously, by extracting features and utilizing redundant data, it reduces upfront investment and data dependence while ensuring high-risk identification accuracy and low false negative rate. This overcomes the bottleneck of traditional evaluation methods where cost and accuracy are difficult to balance, and adapts to the efficient evaluation needs of large-scale charging pile networks. 3. This embodiment provides a reliable method for assessing out-of-tolerance risks. It is an online assessment method for charging pile metering out-of-tolerance risks based on the correlation of error change curves of multiple charging piles at the same station. This method, through a closed-loop architecture design, achieves full-process automation of automatic raw data collection, intelligent calculation of daily average error, feature dimensionality reduction and clustering, automatic risk level determination, and result quantitative verification. The core lies in retaining the stage-specific characteristics of the error change curve through segmented processing, combining PCA dimensionality reduction and K-Means clustering to extract the core trends of multiple charging piles at the same station, and innovatively introducing a three-level classification rule based on trend deviation (normal charging piles, observation charging piles, and high-risk charging piles), forming a continuous monitoring mechanism within the cycle. This method ultimately achieves early identification and precise control of charging pile out-of-tolerance risks under all operating conditions, significantly improving metering management efficiency, avoiding unfair transactions, and ensuring the accuracy of charging settlement and market fairness.

[0024] Preferably, step S1 specifically includes the following steps: S11. Raw metering data collection: Collect the first data from the same station daily at a preset frequency. Measured current value of pile cap Measured voltage value , t Timestamps within each day , To determine the number of data collections per day, in this embodiment, the measured current (unit: A) and measured voltage (unit: V) values ​​of 25 charging piles are collected at six time points each day: 7:00, 11:00, 15:00, 19:00, 21:00, and 23:00, and the corresponding timestamps t=1 to t=6 are recorded. S12. Collection of verification standard values: On the first day of the verification cycle ( d =1), through high-level metering methods, the metering verification of each charging pile is completed, and the first... Current verification standard value of charging piles in Taiwan Voltage verification standard value (This standard value is fixed within the cycle and does not require daily calibration, adapting to different verification cycles.) In this embodiment, on the first day of the cycle (d=1), a standard table that has passed legal verification is used to complete the metrological verification of 25 charging piles, and the current verification standard value and voltage verification standard value of each pile are determined. This standard value is fixed and used within a 90-day cycle. S13. Intraday Error Calculation: Calculate the following formula: i Taiwan charging pile d The first day t Instantaneous measurement error: ; In the formula, For the first i Taiwan charging pile dThe first day t Sub-instantaneous measurement error; In this embodiment, the instantaneous measurement error of each pile is calculated 6 times per day according to the above formula. The measured current value of the 3rd pile on the 10th day (d=10) and the 4th time (t=4) is 85A and the measured voltage value is 380V, which correspond to the standard values ​​of 84.8A and 379.5V. The instantaneous measurement error of that time is calculated by substituting them into the formula. S14. Calculation of daily average error: For the first... d The sky The daily average error is obtained by averaging the instantaneous errors. To avoid interference from intraday high-frequency fluctuations in trend analysis: ; In this embodiment, the average of the six instantaneous errors of each pile per day is taken to obtain the daily average error. The six instantaneous errors of the third pile on the 10th day are 0.2%, 0.3%, 0.25%, 0.28%, 0.32%, and 0.27%, respectively, and its daily average error is 0.27%. S15. Definition of error variation curve: based on the number of days within a single verification cycle. d The horizontal axis represents the average daily error. Using the vertical axis, construct the first i Error variation curve of a charging pile within a single calibration cycle : ; Among them, the error change curve For discrete points, reflecting the first point within the station i The error trend of each charging pile over a single calibration cycle; this embodiment uses the daily average error sequence of each charging pile over 90 days to plot the measurement error variation curves of 25 charging piles at the same station, as shown in the figure. Figure 2 As shown in the figure, the horizontal axis represents the number of days in the calibration cycle (1-90 days), and the vertical axis represents the daily average error (%). The changes in the metering error curves of most charging piles can be clearly observed.

[0025] Preferably, step S2 specifically includes the following steps: S21. Divide the error variation curve of each charging pile into periods proportionally. m The error variation curve of ≥2 consecutive segments is defined as the first segment. p The cumulative number of days corresponding to the segment error change curve is : ; The above constraints are: , DThis refers to the total number of days in a single calibration cycle for a charging pile within the same station. In this embodiment, the metering error variation curve for each charging pile is divided into three consecutive segments according to the cycle ratio, with each segment containing the cumulative number of days. =30 days, integrate the error sequences of each pile for the first 30 days, 31-60 days, and 61-90 days to form three segmented error change curves. The first segment (p=1) curve of the third pile is the daily average error sequence from d=1 to 30 days, the second segment (p=2) is the sequence from d=31 to 60 days, and the third segment (p=3) is the sequence from d=61 to 90 days; When the cumulative number of days At that time, the integration of the first i In front of the charging station Error sequence for days: ; Within the same site i Taiwan charging pile p The error variation curves within each segmented interval are as follows: ; S22. For the error variation curve of each charging pile in the above station, normalization is performed using the following formula to eliminate the impact of differences in the initial error magnitude of different charging piles on the trend comparison: ; In the formula: , They are respectively Minimum and maximum values; based on N Taiwan charging pile p Segment error change curve Construct the high-dimensional feature matrix corresponding to this segment. : ; in, Indicates the first p The number of days included in the segment error variation curve, i.e., the feature dimension; each element in the matrix. Indicates the first The charging station, on the first The first segment error change curve Error value after normalization .

[0026] Preferably, in this embodiment, PCA dimensionality reduction is performed on the high-dimensional matrix of each measurement error change curve segment to ensure the effectiveness and consistency of the stage features. Therefore, step S3 specifically includes the following steps: S31. Standardization of Error Variation Curve Characteristics: For high-dimensional matrices Each column is Z-score standardized to eliminate the interference of variance differences in features across different days on the dimensionality reduction results, ensuring fair contributions of each feature to the principal components and adapting to feature distributions across different testing periods. The formula is as follows: ; In the formula: For the first d The mean of the column features; For the first d Standard deviation of the column features; In this embodiment, Z-Score standardization is performed on the high-dimensional feature matrix of 30 columns for each curve segment, and the mean and standard deviation of each column are calculated. The 10th column of the 1st segment ( d The mean of the data (=10) is 0.25% and the standard deviation is 0.08%. All 25 data points in this column are standardized according to the formula to eliminate variance differences due to different number of days. S32. Covariance Matrix and Eigenvalue Decomposition: List the... p Covariance matrix corresponding to segment error variation curve data : ; The formula for calculating each element is as follows: ; In the formula: Indicates the site N The first charging pile p The first segment of the measurement error change curve a Heaven and the First b The covariance of the error characteristics, with M and O as placeholders, similar to ellipses; For covariance matrix Eigenvalue decomposition yields eigenvalues ​​and eigenvectors: ; Among them, the feature matrix Elements in the matrix Representing the p The first segment of the measurement error change curve The principal component in the first... d Weights on the eigenvalue feature; eigenvalue diagonal matrix , It is the covariance matrix eigenvalues, For the first s The eigenvalues ​​corresponding to the principal components, and satisfying ; Based on the above formula, this embodiment calculates the covariance matrix (30 rows × 30 columns) of the standardized high-dimensional matrix of the first segment. Each element represents the covariance of the error characteristics for the corresponding two days. The element (a=5, b=15) represents the covariance of the error characteristics for the 5th and 15th days of the first segment of the 25 piles. Eigenvalue decomposition is performed on the covariance matrix to obtain 30 eigenvalues ​​and their corresponding eigenvectors. The eigenvalues ​​are sorted from largest to smallest as follows: ; S33, Principal Component Selection: From D p Before selecting from the principal components k (p) One makes the front k (p) The sum of the variance contributions of each principal component accounts for no less than the proportion of the total contribution of all principal components. θ , θ To retain a threshold for information, k (p) = D p , The formula is: ; Combination The above formula can be simplified to: ; In this embodiment, an information retention threshold is set. θ =0.85, choose the smallest one. k (p) This makes the former k (p) The sum of the variance contributions of each principal component must account for no less than 85% of the total contribution of the 30 principal components, as shown in the formula: ; Set information retention threshold θ =0.85, before calculation k (p) The sum of the variance contributions of the first three principal components. Calculations show that the sum of the variance contributions of the first three principal components reaches 86.2%, meeting the requirement of not less than 85%, therefore, they are selected. k (p) =3 principal components; S34. Construction of Low-Dimensional Feature Matrix: Determining k (p) After =3, from the characteristic matrix U (p) Extract the first 3 column vectors to form a 30×3 dimension-reduced matrix. : ; In the formula: elements of the matrix Indicates the first p The first segment error change curve j The principal components after dimensionality reduction are in the... d The weights of the error characteristics of each day are obtained by linear combination. j Low-dimensional features; Standardized high-dimensional matrix With dimensionality reduction matrix Multiply, we get low-dimensional feature matrix : ; low-dimensional feature matrix It can be expanded as follows: ; In the formula, For the first time on the site i Taiwan in p The low-dimensional eigenvector of the segment measurement error variation curve; Indicates the first i The number of charging piles in the first p The first segment of the measurement error change curve j The low-dimensional features are calculated by the following linear combination: .

[0027] Preferably, step S4 specifically includes the following steps: S41. Using the K-Means unsupervised clustering algorithm, all charging piles in the station are clustered. p Charging piles with similar segment measurement error characteristic curves and error change curves are grouped into the same cluster: Optimal number of clusters Determine: The range of the number of clusters C is... (generally , (Adapting to a typical scale of N charging piles); the optimal value is selected through the sum of squared errors within clusters (SSE), the SSE formula is: ; In the formula, For the first c The set of charging station indexes for a cluster. For the first c Cluster center ; The smaller the value, the more consistent the low-dimensional characteristics of the charging piles within the cluster; The number of clusters is calculated using the above formula. C The range is 2≤ C ≤5. WhenC When the SSE value is 2, the minimum SSE value is 0.32. Therefore, the optimal number of clusters is determined. C =2; Execute the K-Means unsupervised clustering algorithm: for low-dimensional feature matrices To perform clustering and avoid local optima caused by random initialization, the steps are as follows: Initialize cluster centers: Select using the K-Means algorithm C= 2 initial centers Ensure that the initial cluster centers are distributed to cover the different error variation curves of the characteristic curve of the p-th segment of the charging piles in the station; Sample allocation: For each low-dimensional feature vector Calculate its relationship with the two cluster centers. Weighted Euclidean distance: ; In the formula, Let be the weight of the j-th feature, and t be the iteration number. Assign to the cluster with the smallest distance In this embodiment, the eigenvector of the third pile is 0.45 away from the center of the first cluster and 0.98 away from the center of the second cluster, so it is assigned to the first cluster. Central update: for each cluster The update center is the mean of the low-dimensional feature vectors of the charging piles within the cluster: ; In the formula, Let be the number of charging piles in the c-th cluster at the t-th iteration; In this embodiment, the mean of all feature vectors in each cluster is calculated. When updating the cluster center, the first cluster contains 22 stakes, and its new center is the mean of the 22 vectors in the cluster. Convergence criterion: Repeated sample assignment and center update until the following condition is met: or reaching the maximum number of iterations. , The convergence threshold is usually set to 1. ; In this embodiment, a convergence threshold is set. After 5 iterations, the change in cluster centers is less than If the convergence condition is met, stop iterating. This embodiment uses the K-Means algorithm to perform cluster analysis on the low-dimensional features of 25 charging piles. The clustering results are as follows: Figure 3 As shown, the cluster distribution and cluster center location of normal charging piles and high-risk charging piles can be intuitively presented, providing a visual basis for subsequent risk level determination; S42. Determination of Out-of-Tolerance Risk Level: Based on the group consistency of the error variation curves of charging piles at the same station during the calibration cycle, and combined with clustering results, the out-of-tolerance risk of each charging pile is determined within the cycle. The risk level classification reflects general coverage of all charging piles, specifically including: Core trend cluster identification: p The sample proportion of the c-th cluster of the segment measurement error variation curve is: ; In the formula, the cluster with the largest proportion is the core trend cluster. ,Right now This represents the changing trend of the mainstream metering error curve of the charging piles at the same station; The sample proportions of the two clusters were calculated. The first cluster had 22 units and the second cluster had 3 units. The first cluster had the largest proportion and was identified as the core trend cluster, representing the mainstream error change curve of the charging piles at the same station.

[0028] Set up risk assessment equations, including: Determining a Normal Charging Pile: A normal charging pile must simultaneously meet the following criteria: belonging to the core trend cluster and being relatively close to the center of the core trend cluster. This ensures that the error variation curve is highly consistent with that of most charging piles. The equation set is as follows: ; In the formula, As the core trend cluster center, Normal trend distance threshold , which is the criterion for Euclidean distance; In this embodiment, a normal device is defined as one that belongs to the core trend cluster and is within 0.5 of the center of the core trend cluster. Among the 22 charging piles in the first cluster, all charging piles are within 0.5 of the center of the core cluster and are therefore considered normal devices. High-risk charging pile judgment equations: A high-risk charging pile meets either the condition that its cluster size is too small or its distance from the center of the core trend cluster is too far, indicating that the error change curve deviates significantly from the group and there is a risk of exceeding the tolerance. The judgment equations are as follows: ; In the formula, The threshold for the proportion of small clusters is 0.1 ≤ r th ≤0.2, The high-risk trend distance threshold is used as the high-risk boundary of the Euclidean distance, where 1.0 ≤ d th2 ≤1.5, and d th2 > d th1 ; In this embodiment, high-risk devices are defined as those that meet the following criteria: the percentage of devices belonging to their respective clusters or the distance from the center of the core trend cluster. In the second cluster, the percentage of 3 devices is 8% ≤ 10%, and one of these devices is 1.05 > 1.0 from the center of the core cluster. A total of 3 devices are identified as high-risk devices. The equations for judging the charging piles to be observed are as follows: The charging pile to be observed must simultaneously meet the following conditions: it does not belong to the core trend cluster, its cluster is not a small cluster, its distance from the center of the core trend cluster is between two thresholds, and it is in a transitional state between normal and high risk. The equations are as follows: ; Based on the calculation results of the risk assessment equation, the risk level label of each charging pile in the same station is finally output, including normal charging piles, observation charging piles and high-risk charging piles; In this embodiment, the observation device does not belong to the core class, its cluster percentage is >10% and its distance from the core center is between 0.5 and 1.0, and no device meets this condition.

[0029] In this embodiment, the risk output is finally carried out: 22 of the 25 charging piles are ultimately determined to be normal devices and 3 are high-risk devices. The risk assessment is completed on the 45th day of the cycle, realizing early warning within the cycle.

[0030] The final output of the out-of-tolerance risk level determination is a risk level label for each charging pile, including three categories: normal charging piles, observation charging piles, and high-risk charging piles. This result intuitively quantifies the degree of deviation between the charging pile error change curve and the mainstream pattern of the group. High-risk charging piles require emergency re-inspection, observation charging piles require continuous monitoring of their trend changes, and normal charging piles are included in the regular inspection cycle. This output provides data support for the hierarchical management of charging pile metering errors, realizing the implementation from full-scale charging pile monitoring to precise control of key charging piles, effectively improving the efficiency of out-of-tolerance hazard identification and the rationality of maintenance resource allocation.

[0031] Preferably, in step S5, calculating the high-risk identification accuracy specifically involves: ; In the formula, The number of charging piles that were predicted to be high-risk and whose deviations were confirmed at the end of the period; This refers to the total number of charging piles identified as high-risk throughout the entire lifecycle of the same station. Acc The target value is: Acc nomal ≥90%; The specific calculation of the missed detection rate of normal charging piles is as follows: ; In the formula: The number of charging piles that were initially considered normal but were confirmed to be out of tolerance at the end of the cycle; The total number of charging piles predicted to be normal throughout the entire cycle; target value .

[0032] In this embodiment, a comprehensive on-site inspection of 25 charging piles was conducted on the 90th day of the cycle (the end of the inspection cycle). The results showed that all 3 high-risk devices were confirmed to be out of tolerance (high-risk identification accuracy rate 100%), while all 22 normal devices were within tolerance (normal device false negative rate 0%). After review, they were included in routine maintenance. All verification indicators met the target requirements, proving that the evaluation results of this invention are reliable.

[0033] This embodiment quantifies and verifies the reliability of the risk assessment results from different dimensions through indicators such as the accuracy rate of high-risk identification and the effectiveness of the missed detection rate of normal charging piles. These indicators link the predicted conclusions with the actual verification data at the end of the cycle, ensuring both the accurate identification of high-risk charging piles and avoiding the missed or misjudged detection of normal charging piles. Ultimately, this provides quantitative support for the scientific and practical nature of the entire risk assessment system, effectively guaranteeing the accuracy of identifying potential hazards of excessive metering errors in charging piles and the rationality of maintenance resource allocation.

[0034] The core of the evaluation method described in the above embodiments of this application is to achieve risk assessment of multiple charging piles exceeding tolerances at the same station through a closed-loop process of data acquisition and preprocessing, PCA dimensionality reduction clustering analysis, and risk assessment result verification. First, measured current and voltage data of charging piles within a single calibration cycle are collected. The daily average error is calculated by combining the standard value at the start of the cycle, and an error variation curve is constructed. After segmentation and standardization, a high-dimensional feature matrix is ​​formed, and then core features are extracted through PCA dimensionality reduction. K-Means clustering is used to identify the core trend clusters of error changes in charging piles at the same station. Finally, based on the cluster proportion and distance threshold, charging piles are divided into three categories: normal, under observation, and high-risk. The accuracy of the assessment is verified by combining the calibration data at the end of the cycle, ultimately achieving early identification and precise control of risk exceeding tolerances within the calibration cycle.

[0035] like Figure 4 As shown, another preferred embodiment of this example provides a device for assessing the risk of multiple charging piles exceeding tolerances at the same station, comprising: The data acquisition and preprocessing module is used to collect measured current and voltage data of multiple charging piles at the same station within a single calibration cycle, obtain calibration standard values ​​through high-precision measurement, calculate intraday instantaneous error and daily average error, and construct an error change curve for a single charging pile. The segmentation and high-dimensional feature matrix construction module divides the error change curve of each charging pile into segments according to the period ratio, eliminates the difference in the initial error magnitude of different charging piles, and constructs the high-dimensional feature matrix corresponding to each segment. The principal component analysis dimensionality reduction module is used to perform PCA dimensionality reduction on high-dimensional feature matrices. It obtains each principal component and its corresponding variance contribution through covariance matrix calculation and eigenvalue decomposition, and extracts core features to construct a low-dimensional feature matrix. The low-dimensional feature-based clustering and out-of-range risk assessment module is used to execute the K-Means unsupervised clustering algorithm based on a low-dimensional feature matrix, and to cluster all charging piles in the station. p Charging piles with similar segment measurement error characteristic curves and error change curves are grouped into the same cluster. The optimal number of clusters and the core trend clusters are determined. Based on the cluster ratio and distance threshold, charging piles are divided into three categories: normal, unobservable, and high-risk. The risk assessment result verification module compares the standard electrical energy value with the actual electrical energy value of the charging pile based on clock synchronization at the end of the verification cycle to obtain the actual verification data of each charging pile. The reliability of the risk assessment result is verified by calculating the high-risk identification accuracy rate and the normal charging pile false detection rate, forming a closed-loop management system. Electrical energy = voltage × current × t.

[0036] The multi-charging pile deviation risk assessment device provided in this embodiment adopts the multi-charging pile deviation risk assessment method in the above embodiments, solving the technical problems of existing charging pile metering performance assessment processes, such as excessive manual intervention and insufficient targeted control leading to slow and inefficient deviation risk identification, and high costs due to reliance on pre-calibrated standard charging piles or large-scale training data. Compared with the prior art, the beneficial effects of the multi-charging pile deviation risk assessment device provided in this embodiment are the same as those of the multi-charging pile deviation risk assessment method provided in the above embodiments, and other technical features in the multi-charging pile deviation risk assessment device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0037] like Figure 5 As shown, a preferred embodiment of this example also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the method for assessing the risk of multiple charging piles exceeding tolerances at the same station as described in the above embodiment.

[0038] This embodiment provides an electronic device that employs the same-site multi-charging-pile out-of-tolerance risk assessment method described in the above embodiments. This addresses the technical problems of existing charging pile metering performance assessment processes, such as excessive manual intervention, insufficient targeted control leading to slow and inefficient out-of-tolerance risk identification, and high costs due to reliance on pre-calibrated standard charging piles or large-scale training data. Compared to existing technologies, the electronic device provided in this embodiment has the same beneficial effects as the same-site multi-charging-pile out-of-tolerance risk assessment method described in the above embodiments, and other technical features of the electronic device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0039] like Figure 6 As shown in the preferred embodiment, this embodiment also provides a computer device, which may be a terminal or a liveness detection server, and its internal structure diagram may be as follows. Figure 6 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with other external computer devices via a network connection. When the computer program is executed by the processor, it implements the steps of the aforementioned method for assessing the risk of multiple charging piles exceeding tolerances at the same station.

[0040] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the solution of this embodiment, and does not constitute a limitation on the computer device to which the solution of this embodiment is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0041] The computer equipment provided in this application employs the same-site multi-charging-pile out-of-tolerance risk assessment method described in the above embodiments. This addresses the technical problems of existing charging pile metering performance assessment processes, which suffer from excessive manual intervention and insufficient targeted control, leading to slow and inefficient out-of-tolerance risk identification, and high costs due to reliance on pre-calibrated standard charging piles or large-scale training data. Compared to existing technologies, the beneficial effects of the computer equipment provided in this embodiment are the same as those of the same-site multi-charging-pile out-of-tolerance risk assessment method provided in the above embodiments. Furthermore, other technical features of the electronic equipment are the same as those disclosed in the methods of the above embodiments, and will not be elaborated upon here.

[0042] A preferred embodiment of this example also provides a storage medium, which includes a stored program that, when the program is executed, controls the device where the storage medium is located to perform the steps of the above-described method for assessing the risk of multiple charging piles exceeding tolerances at the same station.

[0043] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0044] If the functions described in this embodiment are implemented as software functional units and sold or used as independent products, they can be stored in one or more computing device-readable storage media. Based on this understanding, the parts of this embodiment that contribute to the prior art or the technical solution can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a computing device (which may be a personal computer, server, mobile computing device, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this embodiment. The aforementioned storage media include: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0045] Those skilled in the art will understand that the embodiments of this example can be provided as methods, systems, or computer program products. Therefore, this example can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this example can take the form of a computer program product implemented 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. The solutions in this example can be implemented using various computer languages, such as the object-oriented programming language C++ and the embedded programming language C.

[0046] This embodiment is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this embodiment. It should 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, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0047] 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, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0048] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0049] This embodiment also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for assessing the excessive risk of multiple charging piles at the same station.

[0050] The computer program product provided in this embodiment solves the technical problems of existing charging pile metering performance evaluation processes, such as excessive manual intervention, insufficient targeted control leading to slow and inefficient identification of out-of-tolerance risks, and high costs due to reliance on pre-calibrated standard charging piles or large-scale training data. Compared with the prior art, the beneficial effects of the computer program product provided in this embodiment are the same as those of the out-of-tolerance risk assessment method for multiple charging piles at the same station provided in the above embodiments, and will not be repeated here.

[0051] Although preferred embodiments of this embodiment have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this embodiment.

[0052] Obviously, those skilled in the art can make various modifications and variations to this embodiment without departing from the spirit and scope of this embodiment. Therefore, if these modifications and variations of this embodiment fall within the scope of the claims of this embodiment and their equivalents, this embodiment is also intended to include these modifications and variations.

Claims

1. A method for assessing the risk of excessive tolerances in multiple charging piles at the same station, characterized in that, Including the following steps: S1. Collect measured current and voltage data of multiple charging piles at the same station within a single calibration cycle, obtain calibration standard values ​​through high-precision measurement, calculate the instantaneous error and daily average error, and construct the error change curve of a single charging pile; S2. Divide the error change curve of each charging pile into segments according to the period ratio, eliminate the difference in the initial error magnitude of different charging piles, and construct the high-dimensional feature matrix corresponding to each segment. S3. The high-dimensional feature matrix of each error change curve is reduced by PCA. The principal components and their corresponding variance contributions are obtained by covariance matrix calculation and eigenvalue decomposition. Core features are extracted to construct a low-dimensional feature matrix. S4. Execute the K-Means unsupervised clustering algorithm based on the low-dimensional feature matrix to cluster all charging piles within the station. p Charging piles with similar segment measurement error characteristic curves and error change curves are grouped into the same cluster. The optimal number of clusters and the core trend clusters are determined. Based on the cluster ratio and distance threshold, charging piles are divided into three categories: normal, unobservable, and high-risk. S5. At the end of the verification cycle, based on clock synchronization, the verification standard power value is compared with the actual measured power value of the charging pile to obtain the actual verification data of each charging pile. The reliability of the risk assessment results is verified by calculating the high-risk identification accuracy rate and the normal charging pile false detection rate, thus forming a closed-loop management system.

2. The method for assessing the risk of multiple charging piles exceeding tolerances at the same station according to claim 1, characterized in that, Step S1 specifically includes the following steps: S11. Raw metering data collection: Collect the first data from the same station daily at a preset frequency. Measured current value of pile cap Measured voltage value , t Timestamps within each day , This refers to the number of times data is collected per day. S12. Verification Standard Value Collection: On the starting date of the verification cycle, the standard value is determined using a high-level metrological method. Current verification standard value of charging piles in Taiwan Voltage verification standard value ; S13. Intraday Error Calculation: Calculate the following formula: i Taiwan charging pile d The first day t Instantaneous measurement error: ； In the formula, For the first i Taiwan charging pile d The first day t Sub-instantaneous measurement error; S14. Calculation of daily average error: For the first... d The sky The daily average error is obtained by averaging the instantaneous errors. To avoid interference from intraday high-frequency fluctuations in trend analysis: ； S15. Definition of error variation curve: based on the number of days within a single verification cycle. d The horizontal axis represents the average daily error. Using the vertical axis, construct the first i Error variation curve of a charging pile within a single calibration cycle : ； Among them, the error change curve For discrete points, reflecting the first point within the station i The error trend of the charging pile over time within a single calibration cycle.

3. The method for assessing the risk of multiple charging piles exceeding tolerances at the same station according to claim 1, characterized in that, Step S2 specifically includes the following steps: S21. Divide the error variation curve of each charging pile into periods proportionally. m The error variation curve of ≥2 consecutive segments is defined as the first segment. p The cumulative number of days corresponding to the segment error change curve is : ； The above constraints are: , D The total number of days in the calibration cycle for a single charging pile within the same station; When the cumulative number of days At that time, the integration of the first i In front of the charging station Error sequence for days: ； Within the same site i Taiwan charging pile p The error variation curves within each segmented interval are as follows: ； S22. For the error variation curve of each charging pile in the above station, normalization is performed using the following formula to eliminate the impact of differences in the initial error magnitude of different charging piles on the trend comparison: ； In the formula: , They are respectively Minimum and maximum values; based on N Taiwan charging pile p Segment error change curve Construct the high-dimensional feature matrix corresponding to this segment. : ； in, Indicates the first p The number of days included in the segment error variation curve, i.e., the feature dimension; each element in the matrix. Indicates the first The charging station, on the first The first segment error change curve Error value after normalization .

4. The method for assessing the risk of multiple charging piles exceeding tolerances at the same station according to claim 1, characterized in that, Step S3 specifically includes the following steps: S31. Standardization of Error Variation Curve Characteristics: For high-dimensional matrices Each column is Z-score standardized to eliminate the interference of variance differences in features across different days on the dimensionality reduction results, ensuring fair contributions of each feature to the principal components and adapting to feature distributions across different testing periods. The formula is as follows: ； In the formula: For the first d The mean of the column features; For the first d Standard deviation of the column features; S32. Covariance Matrix and Eigenvalue Decomposition: List the... p Covariance matrix corresponding to segment error variation curve data : ； The formula for calculating each element is as follows: ； In the formula: Indicates the site N The first charging pile p The first segment of the measurement error change curve a Heaven and the First b Covariance of the error characteristics; For covariance matrix Eigenvalue decomposition yields eigenvalues ​​and eigenvectors: ； Among them, the feature matrix Elements in the matrix Representing the p The first segment of the measurement error change curve The principal component in the first... d Weights on the eigenvalue feature; eigenvalue diagonal matrix , It is the covariance matrix eigenvalues, For the first s The eigenvalues ​​corresponding to the principal components, and satisfying ; S33, Principal Component Selection: From D p Before selecting from the principal components k (p) One makes the front k (p) The sum of the variance contributions of each principal component accounts for no less than the proportion of the total contribution of all principal components. , To retain a threshold for information, k (p) = D p , The formula is: ； Combination The above formula can be simplified to: ； S34. Construction of Low-Dimensional Feature Matrix: Determining k (p) Then, from the feature matrix U (p) Extracting the first k (p) column vectors constitute Dimensionality reduction matrix : ； In the formula: elements of the matrix Indicates the first p The first segment error change curve j The principal components after dimensionality reduction are in the... d The weights of the error characteristics of each day are obtained by linear combination. j Low-dimensional features; Standardized high-dimensional matrix With dimensionality reduction matrix Multiply, we get low-dimensional feature matrix : ； low-dimensional feature matrix It can be expanded as follows: ； In the formula, For the first time on the site i Taiwan in p The low-dimensional eigenvector of the segment measurement error variation curve; Indicates the first i The number of charging piles in the first p The first segment of the measurement error change curve j The low-dimensional features are calculated by the following linear combination: .

5. The method for assessing the risk of multiple charging piles exceeding tolerances at the same station according to claim 4, characterized in that, In step S33, from D p Before selecting from the principal components k (p) When there are multiple options, choose the smallest one. k (p) ,1≤ k (p) ≤ D p Make the former k (p) The sum of the variance contributions of each principal component accounts for no less than the proportion of the total contribution of all principal components. θ When θ is taken to be .

6. The method for assessing the risk of multiple charging piles exceeding tolerances at the same station according to claim 1, characterized in that, Step S4 specifically includes the following steps: S41. Using the K-Means unsupervised clustering algorithm, all charging piles in the station are clustered. p Charging piles with similar segment measurement error characteristic curves and error change curves are grouped into the same cluster: Optimal number of clusters Determine: The range of the number of clusters C is... , , This is suitable for a typical scale of N charging piles; the optimal value is selected through the sum of squared errors within clusters (SSE), and the SSE formula is: ； In the formula, For the first c The set of charging station indexes for a cluster. For the first c Cluster centers; The smaller the value, the more consistent the low-dimensional characteristics of the charging piles within the cluster; Execute the K-Means unsupervised clustering algorithm: for low-dimensional feature matrices To perform clustering and avoid local optima caused by random initialization, the steps are as follows: Initialize cluster centers: Select using the K-Means algorithm C Initial center Ensure that the initial cluster centers are distributed to cover the different error variation curves of the characteristic curve of the p-th segment of the charging piles in the station; Sample allocation: For each low-dimensional feature vector Calculate its relationship with each cluster center Weighted Euclidean distance: ； In the formula, Let be the weight of the j-th feature, and t be the iteration number. Assign to the cluster with the smallest distance ; For each cluster The update center is the mean of the low-dimensional feature vectors of the charging piles within the cluster: ； In the formula, Let be the number of charging piles in the c-th cluster at the t-th iteration; Convergence criterion: Repeated sample assignment and center update until the following condition is met: or reaching the maximum number of iterations. , The convergence threshold is usually set to 1. ; S42. Determination of Out-of-Tolerance Risk Level: Based on the group consistency of the error variation curves of charging piles at the same station during the calibration cycle, and combined with clustering results, the out-of-tolerance risk of each charging pile is determined within the cycle. The risk level classification reflects general coverage of all charging piles, specifically including: Core trend cluster identification: p The sample proportion of the c-th cluster of the segment measurement error variation curve is: ； In the formula, the cluster with the largest proportion is the core trend cluster. ,Right now This represents the changing trend of the mainstream metering error curve of the charging piles at the same station; Set up risk assessment equations, including: Determining a Normal Charging Pile: A normal charging pile must simultaneously meet the following criteria: belonging to the core trend cluster and being relatively close to the center of the core trend cluster. This ensures that the error variation curve is highly consistent with that of most charging piles. The equation set is as follows: ； In the formula, As the core trend cluster center, Normal trend distance threshold , which is the criterion for Euclidean distance; High-risk charging pile judgment equations: A high-risk charging pile meets either the condition that its cluster size is too small or its distance from the center of the core trend cluster is too far, indicating that the error change curve deviates significantly from the group and there is a risk of exceeding the tolerance. The judgment equations are as follows: ； In the formula, The threshold for the proportion of small clusters is 0.1 ≤ r th ≤0.2, The high-risk trend distance threshold is used as the high-risk boundary of the Euclidean distance, where 1.0 ≤ d th2 ≤1.5, and d th2 > d th1 ; The equations for judging the charging piles to be observed are as follows: The charging pile to be observed must simultaneously meet the following conditions: it does not belong to the core trend cluster, its cluster is not a small cluster, its distance from the center of the core trend cluster is between two thresholds, and it is in a transitional state between normal and high risk. The equations are as follows: ； Based on the risk assessment equation, the final output is the risk level label for each charging pile at the same station, including normal charging piles, observation charging piles, and high-risk charging piles.

7. The method for assessing the risk of multiple charging piles exceeding tolerances at the same station according to claim 1, characterized in that, In step S5, the calculation of the high-risk identification accuracy rate is specifically as follows: ； In the formula, The number of charging piles that were predicted to be high-risk and whose deviations were confirmed at the end of the period; This refers to the total number of charging piles identified as high-risk throughout the entire lifecycle of the same station. Acc The target value is: Acc nomal ≥90%; The specific calculation of the missed detection rate of normal charging piles is as follows: ； In the formula: The number of charging piles that were initially considered normal but were confirmed to be out of tolerance at the end of the cycle; The total number of charging piles predicted to be normal throughout the entire cycle; target value .

8. A device for assessing the risk of multiple charging piles exceeding tolerances at the same station, characterized in that, include: The data acquisition and preprocessing module is used to collect measured data of current, voltage, and energy of multiple charging piles at the same station within a single calibration cycle, obtain calibration standard values ​​through high-precision measurement, calculate intraday instantaneous error and daily average error, and construct an error change curve for a single charging pile. The segmentation and high-dimensional feature matrix construction module divides the error change curve of each charging pile into segments according to the period ratio, eliminates the difference in the initial error magnitude of different charging piles, and constructs the high-dimensional feature matrix corresponding to each segment. The principal component analysis dimensionality reduction module is used to perform PCA dimensionality reduction on high-dimensional feature matrices. It obtains each principal component and its corresponding variance contribution through covariance matrix calculation and eigenvalue decomposition, and extracts core features to construct a low-dimensional feature matrix. The low-dimensional feature-based clustering and out-of-range risk assessment module is used to execute the K-Means unsupervised clustering algorithm based on a low-dimensional feature matrix, and to cluster all charging piles in the station. p Charging piles with similar segment measurement error characteristic curves and error change curves are grouped into the same cluster. The optimal number of clusters and the core trend clusters are determined. Based on the cluster ratio and distance threshold, charging piles are divided into three categories: normal, unobservable, and high-risk. The risk assessment result verification module compares the standard power value with the actual power value of the charging pile based on clock synchronization at the end of the verification cycle to obtain the actual verification data of each charging pile. The reliability of the risk assessment results is verified by calculating the high-risk identification accuracy rate and the normal charging pile false negative rate, thus forming a closed-loop management system.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for assessing the risk of excessive charges at the same station for multiple charging piles as described in any one of claims 1 to 7.

10. A storage medium comprising a stored program, characterized in that, When the program is running, it controls the device where the storage medium is located to perform the steps of the method for assessing the risk of excessive deviation of multiple charging piles at the same station as described in any one of claims 1 to 7.

Images