Vehicle pile network cooperative charging facility abnormal state detection method

Abstract

The invention, which relates to the technical field of charging facility management, discloses a method for detecting an abnormal state of a charging facility based on vehicle pile network cooperation, and the method comprises the steps: reading real-time cooperation information, carrying out the abnormality detection, and building a first abnormality recognition result; performing long and short sequence power grid abnormity identification, and establishing a second abnormity identification result; performing joint influence analysis on the first and second anomaly recognition results to generate a third anomaly recognition result; and performing time sequence change analysis on the operation data information of the charging pile, and performing anomaly fusion analysis on the time sequence change analysis and the third anomaly identification result to generate an abnormal state detection result. The technical problem that a traditional charging facility anomaly detection method in the prior art focuses on analysis of a single data source and is low in detection accuracy and timeliness is solved, multi-level anomaly recognition and elimination analysis are carried out by combining data of the vehicle, the charging pile and the power grid, and the detection accuracy and timeliness are improved. And the accuracy and timeliness of abnormal state detection of the charging facility are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of charging facility management, and particularly to a method for detecting abnormal states of charging facilities with vehicle-pile-network collaboration. Background Art

[0002] With the popularization of electric vehicles and the rapid development of charging facilities, the stability and reliability of charging facilities have become key factors in ensuring the healthy development of the electric vehicle industry. However, during the actual operation of charging facilities, they are often affected by various factors. For example, abnormal behaviors during vehicle charging, mechanical or electrical failures of charging piles themselves, and power grid fluctuations may all lead to abnormal states of charging facilities, thereby affecting user experience and charging safety.

[0003] Currently, traditional methods for detecting abnormal states of charging facilities often focus on the analysis of a single data source, such as only relying on the operation data of charging piles or the charging status of vehicles. This method is often difficult to comprehensively and accurately identify abnormalities in a complex and changeable actual environment, and the accuracy and timeliness of detection are not good. Summary of the Invention

[0004] This application provides a method for detecting abnormal states of charging facilities with vehicle-pile-network collaboration, which is used to solve the technical problem that traditional methods for detecting abnormal states of charging facilities in the prior art mostly focus on the analysis of a single data source, and the accuracy and timeliness of detection are relatively low.

[0005] This application provides a method for detecting abnormal states of charging facilities with vehicle-pile-network collaboration. The method includes: reading real-time collaboration information, where the real-time collaboration information includes vehicle status data information, charging pile operation data information, and power grid monitoring data information, and the real-time collaboration information is information aligned by time stamps; performing data parsing on the real-time collaboration information, and configuring an abnormal recognition channel matching the vehicle according to the parsing result, and performing abnormal detection on the vehicle based on the abnormal recognition channel to establish a first abnormal recognition result; invoking power grid retrospective data based on the power grid monitoring data information, and performing long and short sequence power grid abnormal recognition based on the power grid retrospective data and the power grid monitoring data information to establish a second abnormal recognition result; establishing a joint abnormal exclusion rule, and performing joint influence analysis on the first abnormal recognition result and the second abnormal recognition result based on the joint abnormal exclusion rule to establish compensated residual data; using the compensated residual data and the charging pile operation data information to perform residual abnormal detection on the charging pile to generate a third abnormal recognition result; configuring a sliding window, and performing time series change analysis on the charging pile operation data information based on the sliding window to generate a fourth abnormal recognition result; performing abnormal fusion analysis on the third abnormal recognition result and the fourth abnormal recognition result to generate an abnormal state detection result of the charging pile.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] The method for detecting abnormal states of charging facilities with vehicle-pile-network collaboration provided in this application relates to the technical field of charging facility management. It detects abnormal vehicle states by reading real-time collaboration information, establishes a first abnormal recognition result, conducts abnormal recognition of the power grid for long and short sequences, establishes a second abnormal recognition result, conducts residual abnormal detection through the joint impact analysis of the first and second abnormal recognition results, generates a third abnormal recognition result, and conducts time series change analysis of the charging pile operation data information to generate a fourth abnormal recognition result. The fourth abnormal recognition result is fused with the third abnormal recognition result to generate an abnormal state detection result, solving the technical problem in the prior art that traditional charging facility abnormal detection methods mostly focus on the analysis of a single data source, with low accuracy and timeliness of detection. It realizes the technical effect of combining vehicle, charging pile, and power grid data to conduct multi-level abnormal recognition and exclusion analysis, improving the accuracy and timeliness of charging facility abnormal state detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0009] Figure 1 It is a schematic flowchart of the method for detecting abnormal states of charging facilities with vehicle-pile-network collaboration provided in the embodiments of this application;

[0010] Figure 2 It is a schematic flowchart of configuring a sliding window in the method for detecting abnormal states of charging facilities with vehicle-pile-network collaboration provided in the embodiments of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0011] This application provides a method for detecting abnormal states of charging facilities with vehicle-pile-network collaboration, which is used to solve the technical problem in the prior art that traditional charging facility abnormal detection methods mostly focus on the analysis of a single data source, with low accuracy and timeliness of detection.

[0012] The following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the drawings in the embodiments of this application. Obviously, the described embodiments are only some embodiments of this application, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts belong to the scope of protection of this application.

[0013] It should be noted that the terms "first", "second", etc. in the description of this application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of this application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0014] Embodiment, such as Figure 1 As shown, this application provides a method for detecting abnormal states of charging facilities in vehicle-pile-network collaboration. The method includes:

[0015] P10: Read real-time collaboration information, where the real-time collaboration information includes vehicle status data information, charging pile operation data information, and power grid monitoring data information, and the real-time collaboration information is information aligned by time stamps.

[0016] Specifically, read real-time collaboration information from multiple data sources. The real-time collaboration information includes vehicle status data information, charging pile operation data information, and power grid monitoring data information. Among them, the vehicle status data information refers to various data related to the current operating state of the vehicle, including but not limited to battery power, charging demand, voltage and current parameters, location data, and the connection status between the vehicle and the charging pile. These data are collected in real time through on-vehicle sensors and control systems and can reflect the immediate state of the vehicle during the charging process.

[0017] The charging pile operation data information relates to the working conditions of the charging pile, including the real-time output power, current and voltage parameters, charging rate, working temperature, and the connection status between the charging pile and the power grid of the charging pile. These data are usually obtained through the built-in monitoring system and communication module of the charging pile and can reflect the operating state of the charging pile in real time. The power grid monitoring data information covers the real-time operating state of the power grid, including voltage fluctuations, frequency changes, load conditions, power quality, etc. These information are provided by the monitoring system of the power grid (such as the SCADA system) and can reflect the power supply capacity and stability of the current power grid to the charging pile.

[0018] Moreover, the above real-time collaborative information is aligned by timestamps, ensuring the synchronization of various types of data in the time dimension and facilitating subsequent data parsing and anomaly detection. Timestamp alignment refers to the unified processing of time tags from different data sources to ensure that data from different sources can be compared and analyzed at the same time point. This processing method can avoid analysis biases caused by time differences in data, thereby improving the accuracy of overall anomaly detection.

[0019] Through this series of operations, the system can obtain comprehensive and synchronized real-time status data of vehicles, charging piles, and the power grid, providing a solid data foundation for subsequent anomaly detection steps.

[0020] P20: Parse the real-time collaborative information, configure an anomaly recognition channel matching the vehicle based on the parsing results, perform anomaly detection on the vehicle based on the anomaly recognition channel, and establish a first anomaly recognition result.

[0021] Optionally, conduct in-depth data parsing on the real-time collaborative information, including a detailed combing and interpretation of vehicle status data information, charging pile operation data information, and power grid monitoring data information, aiming to extract key information valuable for vehicle anomaly detection from the massive data. Specifically, first use data parsing techniques such as regular expression matching, JSON / XML parsing, etc. to perform structured processing on the received real-time collaborative information. This step converts the original data into a format easy to analyze and process, such as key-value pairs, tables, or objects. Through data parsing, the system can clearly identify the meanings and interrelationships of each data item, laying a foundation for subsequent steps.

[0022] Furthermore, based on the results of data parsing, dynamically configure an anomaly recognition channel matching the current vehicle according to the specific type, model, charging requirements, etc. of the vehicle. The anomaly recognition channel is a collection of a series of predefined rules, algorithms, or models that are optimized for different types of vehicles and potential anomaly scenarios. For example, for electric vehicles, the system may configure recognition channels for problems such as excessive battery temperature, abnormal charging speed, and charging interface failure.

[0023] After configuring the anomaly recognition channel, perform anomaly detection on the vehicle based on these channels. This includes using various technical means such as data comparison, threshold judgment, and pattern recognition to compare real-time data with preset thresholds, standard patterns, or historical data to identify abnormal situations deviating from the normal range. Generate a first anomaly recognition result according to the results of anomaly detection. This result may include information such as specific anomaly types, the time when the anomaly occurred, and the scope of influence. It is not only an important input for subsequent steps (such as power grid anomaly recognition, charging pile residual anomaly detection, etc.), but also an important basis for users to understand the vehicle status and take timely countermeasures.

[0024] P30: Invoke grid backtracking data based on the grid monitoring data information, perform grid anomaly identification for long and short sequences based on the grid backtracking data and grid monitoring data information, and establish a second anomaly identification result.

[0025] It should be understood that based on the previously obtained grid monitoring data information, the grid backtracking data is invoked for in-depth grid anomaly identification analysis. Among them, the grid monitoring data information serves as the source of real-time data, and the grid monitoring data information provides the operating status of the current grid, including key parameters such as voltage, current, frequency, power factor, etc. These data are collected in real time by grid monitoring devices (such as smart meters, sensors, etc.) and transmitted to the system, providing the basic data for subsequent anomaly identification.

[0026] Different from real-time data, grid backtracking data is a collection of historical data, which records the operating status and change trends of the grid over a past period of time. These data are of great significance for analyzing the long-term stability of the grid and identifying potential fault patterns. The system invokes the grid backtracking database to obtain historical data related to the current monitoring time point for more comprehensive analysis.

[0027] In order to more accurately identify grid anomalies, the long and short sequence analysis technology is adopted to perform grid anomaly identification for long and short sequences based on the grid backtracking data and grid monitoring data information. By combining the advantages of long-term trend analysis and short-term fluctuation monitoring, both the long-term stability and short-term mutability in the grid operation are captured. For example, first, use long-term sequence data (such as data in the past few hours, days, or weeks) to analyze the basic operation mode and change trend of the grid; then, combine short-term sequence data (such as data in the current few minutes or seconds) to monitor the real-time fluctuations and anomalies of the grid.

[0028] Based on the results of the long and short sequence analysis, use a preset anomaly identification algorithm or model to comprehensively evaluate the grid state. These algorithms or models may include statistical methods, machine learning algorithms (such as support vector machines, random forests, etc.) or deep learning models (such as convolutional neural networks, recurrent neural networks, etc.). By comparing the deviation of real-time data from the normal mode, analyzing the similarity of historical data, etc., identify anomalies in the grid, such as voltage dips, frequency offsets, power supply interruptions, etc.

[0029] Finally, based on the identified grid anomaly phenomena, establish a second anomaly identification result. This result contains the anomaly information of the grid within a specific time period, specifically including the anomaly type, occurrence frequency, influence range, etc., and will provide key data support for the joint anomaly analysis in the subsequent steps to ensure the comprehensiveness and accuracy of the anomalies.

[0030] P40: Establish a combined anomaly exclusion rule, and conduct a combined impact analysis of the first anomaly recognition result and the second anomaly recognition result based on the combined anomaly exclusion rule to establish compensated residual data.

[0031] In a possible embodiment of the present application, it is necessary to establish a combined anomaly exclusion rule, and conduct an in-depth combined impact analysis on the first anomaly recognition result (mainly focusing on the vehicle state) and the second anomaly recognition result (focusing on the grid state) based on these rules. This process aims to eliminate false alarms that may be caused by single-system anomalies, while identifying complex anomaly situations of the interaction between systems, and then generating more accurate anomaly detection results.

[0032] Among them, the combined anomaly exclusion rule is formulated based on an in-depth understanding of the interaction mechanism among the vehicle, the charging pile, and the grid, and may include a series of logical judgment conditions, priority rankings, causal relationship chains, etc., for evaluating the relevance, independence, and possible mutual influence among different anomalies. For example, when detecting an abnormal vehicle charging speed, the system may first check whether there is an abnormal grid voltage fluctuation at the same time, because there may be a direct causal relationship between the two.

[0033] After establishing the combined anomaly exclusion rule, the system starts to apply these rules and conduct a comprehensive analysis of the first and second anomaly recognition results. This process involves various technical means such as data fusion, pattern matching, and causal reasoning. The system will evaluate the mutual influence among each anomaly one by one according to the logical relationship and priority defined in the rule, and identify which anomalies are independent, which are caused by other anomalies, and which are complex anomaly situations that need further attention.

[0034] Furthermore, based on the combined impact analysis, compensated residual data is generated. These data are the results of correcting and compensating the original anomaly recognition results, aiming to eliminate false alarms or duplicate reports caused by the interaction between systems. The compensated residual data may include a re-evaluation of the anomaly severity, a fine-tuning of the anomaly occurrence time, a more accurate judgment of the anomaly cause, etc., providing more accurate and reliable input for subsequent charging pile anomaly detection.

[0035] To implement the above process, the system needs to rely on a series of advanced technical supports. First, data fusion technology can integrate information from different data sources to form a unified analysis view. Second, pattern matching and causal reasoning algorithms can help the system identify complex anomaly patterns and causal relationship chains. In addition, machine learning technology can also be applied to this process to automatically learn and identify the relevance and mutual influence among anomalies by training models.

[0036] P50: Use the compensated residual data and the charging pile operation data information to detect the residual anomalies of the charging pile and generate a third anomaly recognition result.

[0037] Specifically, use the compensated residual data generated in the previous step and combine it with the charging pile operation data information to detect the residual anomalies of the charging pile. Among them, the compensated residual data is the abnormal information retained after being processed by the joint anomaly exclusion rule. These data represent the abnormal signals that still exist after excluding the possible influences of the vehicle and the power grid. Therefore, these residual data can more directly reflect the problems of the charging pile itself. And the charging pile operation data information includes various real-time parameters during the operation of the charging pile, such as current, voltage, temperature, power output, etc. These data can directly reflect the working state of the charging pile and help identify possible abnormal situations.

[0038] During the residual anomaly detection process, comprehensively analyze the compensated residual data and the charging pile operation data information. Use the residual anomaly detection technology to compare and analyze the compensated residual data and the charging pile operation data information. This process may involve various technical means such as data comparison, threshold judgment, and trend analysis. According to the preset anomaly detection algorithm or model, evaluate each performance index of the charging pile one by one to identify the abnormal situations with significant deviations from the normal state.

[0039] The advantage of this residual analysis is that it can exclude the interference of external environmental factors (such as vehicle or grid anomalies) and more accurately focus on the operation state of the charging pile itself. Through this method, the system can detect internal problems that may affect the normal operation of the charging pile, such as charging pile hardware failures, sensor malfunctions, software anomalies, etc. And finally generate a third anomaly recognition result. This result specifically describes the abnormal state of the charging pile, including the anomaly type, the affected range, and the possible failure reasons. This provides clear guidance for subsequent maintenance, repair, and further anomaly handling, ensuring that the charging pile can be repaired and adjusted in time when problems occur, thus guaranteeing its normal operation.

[0040] P60: Configure a sliding window and perform time series change analysis on the charging pile operation data information based on the sliding window to generate a fourth anomaly recognition result.

[0041] Further, as Figure 2 shown, for the configuration of the sliding window, step P60 of this application embodiment further includes:

[0042] P61: Use the charging pile operation data information as input data and input it into the data trend fluctuation recognition model; P62: Obtain the fluctuation recognition result of the data trend fluctuation recognition model, use the fluctuation recognition result as matching data, perform data matching of the window mapping sequence, and generate the size constraint of the sliding window; P63: Configure the sliding window according to the size constraint of the sliding window.

[0043] It should be understood that a sliding window is configured, and based on this sliding window, time series change analysis is performed on the operation data information of the charging pile, so as to generate the fourth anomaly recognition result. The sliding window is a commonly used technique in time series analysis. It can analyze the data change trend within a dynamic time range and help identify possible abnormal situations.

[0044] Among them, the configuration process of the sliding window can be, first, use the charging pile operation data information as input data and send these data into the data trend fluctuation recognition model. This model is a pre-trained algorithm or machine learning model that can identify trend changes and fluctuation patterns in the data, including periodic fluctuations, trend growth or decline, sudden anomalies, etc. This kind of model is usually implemented based on time series analysis, statistical methods or machine learning algorithms and can accurately identify abnormal trends in the data.

[0045] Furthermore, according to the output result of the data trend fluctuation recognition model, that is, the fluctuation recognition result, to determine the fluctuation range and change frequency in the data. The fluctuation recognition result is a specific description of the data trend and fluctuation by the model, usually including information such as the amplitude, periodicity, and frequency of the fluctuation. Use the fluctuation recognition result to perform data matching of the window mapping sequence, that is, according to the fluctuation characteristics of the data, match a suitable sliding window size. The size constraint of the sliding window refers to the time range or data point limit of the sliding window, which ensures that the sliding window can effectively cover the key change areas in the data, so as to accurately capture anomalies.

[0046] Finally, configure the sliding window according to the size constraint determined in the previous step. The configured sliding window will move on the time series data and perform local data analysis at each time step. This kind of analysis can identify sudden anomalies and trend changes in the data, so as to generate the fourth anomaly recognition result.

[0047] The fourth anomaly recognition result is the output based on the sliding window analysis, which can reflect the abnormal conditions of the charging pile in different time periods, especially the potential problems that gradually appear over time. Through time series change analysis, the system can more comprehensively master the operation status of the charging pile, early warning of possible development trend problems, and thus take maintenance or adjustment measures in time to ensure the long-term stable operation of the charging pile.

[0048] Further, step P60 of the embodiment of the present application further includes:

[0049] P64: Perform time series change analysis on the charging pile operation data information through the formula as follows:

[0050] Wherein, SW(t) represents the sliding window score of the operation parameter calculated at time point t, W is the size of the sliding window, t is the time point, i represents the current loop index, and F c (i) represents the eigenvalue of the charging pile operation data at time point i, and F c (i - 1) represents the eigenvalue of the charging pile operation data at time point i - 1.

[0051] Optionally, perform time series change analysis on the charging pile operation data information based on the sliding window, that is, perform time series change analysis on the charging pile operation data information through the given formula to obtain a more accurate fourth anomaly detection result. The formula is as follows:

[0052] Wherein, SW(t) represents the sliding window score of the operation parameter calculated at time point t. The sliding window score is an index calculated based on the fluctuation of the charging pile operation data within a period of time, reflecting the stability or fluctuation degree of the data within this period. W is the size of the sliding window, that is, the time range covered by the sliding window or the number of data points. The size of the sliding window directly affects the calculation range of the score. The larger the window, the more historical data the score will consider, thus capturing longer-term trend changes; on the contrary, a smaller window pays more attention to short-term fluctuations. t is the current time point. In the formula, time point t is the end point of the sliding window, and the data change within the time period from t - W + 1 to t is calculated. i represents the current loop index, which starts to loop from time point t - W + 1 until the current time point t. Each i corresponds to the data within a time point. F c (i) represents the eigenvalue of the charging pile operation data at time point i, and F c (i - 1) represents the eigenvalue of the charging pile operation data at time point i - 1. These eigenvalues are usually the key operation parameters of the charging pile, such as voltage, current or power, etc., representing the operation state of the charging pile at different time points.

[0053] Through the calculation of this formula, the system can quantify the degree of change of the time series data within the sliding window. If the sliding window score is high, it indicates that the charging pile operation data fluctuates greatly within this period, and there may be anomalies or instability; on the contrary, a low score indicates that the data is relatively stable.

[0054] P70: Perform abnormal fusion analysis on the third abnormal recognition result and the fourth abnormal recognition result to generate the abnormal state detection result of the charging pile.

[0055] Further, step P70 of the application embodiment further includes:

[0056] P71: Configure a high-confidence abnormal discrimination network, and the high-confidence abnormal discrimination network is as follows:

[0057] Among them, S(t) represents the comprehensive abnormal score, R(t) is the abnormal score of the third recognized abnormal result, T(t) is the abnormal score of the fourth abnormal recognition result, and θ R is the threshold of the residual abnormal score, and θ T is the threshold of the time series abnormal score, w R represents the weight of the residual abnormal evaluation, and w T represents the weight of the time series abnormal evaluation. represents an indicator function for conditional judgment. When the condition in the parentheses is true, otherwise P72: Complete the abnormal fusion analysis according to the high-confidence abnormal discrimination network.

[0058] It should be understood that abnormal fusion analysis is performed on the third abnormal recognition result and the fourth abnormal recognition result, and finally the abnormal state detection result of the charging pile is generated. Abnormal fusion analysis refers to combining multiple abnormal recognition results to form a more comprehensive and accurate abnormal discrimination result, so as to reduce false alarms and improve the reliability of detection.

[0059] To achieve this goal, the system first configures a high-confidence abnormal discrimination network, and the calculation formula of this discrimination network is as follows:

[0060] Among them, S(t) represents the comprehensive abnormal score, which is the total score after fusing and analyzing the third and fourth abnormal recognition results. This score reflects the overall degree of possible abnormality of the charging pile at time point t. R(t) is the abnormal score of the third recognized abnormal result, T(t) is the abnormal score of the fourth abnormal recognition result, and θ R is the threshold of the residual abnormal score, and θ T is the threshold of the time series abnormal score. These thresholds are set according to historical data and actual application scenarios and are used to distinguish normal and abnormal states. w R represents the weight of the residual abnormal evaluation, and w T represents the weight of the time series abnormal evaluation. These weights are used to balance the influence of different abnormal recognition results in the comprehensive abnormal score and are adjusted according to different scenarios or requirements. A characterization indicator function for conditional judgment. When the condition in the brackets is true, otherwise This function is used in the formula to determine whether the current anomaly score exceeds the threshold and adjust the score calculation method accordingly.

[0061] Based on the above high-confidence anomaly discrimination network, anomaly fusion analysis is completed. By comprehensively considering the third and fourth anomaly recognition results and applying the indicator function for conditional judgment, the system can more accurately evaluate whether the operating state of the charging pile is abnormal. If the comprehensive anomaly score S(t) exceeds a preset threshold, the system will determine that there is an abnormal situation with the charging pile and generate corresponding alarm or processing instructions.

[0062] Through the above high-confidence anomaly discrimination network, the system can effectively reduce the false alarm rate, improve the accuracy of anomaly detection, ensure that the charging pile can be detected and processed in a timely manner when an anomaly occurs, and ensure the safe and stable operation of the equipment.

[0063] Furthermore, step P72 of the application embodiment further includes:

[0064] P72-1: Configure the potential anomaly recognition gain coefficient. The activation condition of the potential anomaly recognition gain coefficient is T(t)≥θ T , and R(t)<θ R ; P72-2: Perform high-confidence anomaly discrimination network compensation according to the potential anomaly recognition gain coefficient to complete anomaly fusion analysis.

[0065] Specifically, in order to improve the accuracy and sensitivity of anomaly detection, a potential anomaly recognition gain coefficient is introduced to enhance the recognition ability of potential anomalies under specific conditions, thereby optimizing the performance of the high-confidence anomaly discrimination network.

[0066] First, configure the potential anomaly recognition gain coefficient. The activation condition of this coefficient is: at time point t, the anomaly score T(t) of the fourth anomaly recognition result exceeds its threshold θ T , that is, T(t)≥θ T ; at the same time, the anomaly score R(t) of the third anomaly recognition result is lower than its threshold θ R , that is, R(t)<θ R . When these conditions are met simultaneously, the system will activate the potential anomaly recognition gain coefficient. The logic of activating this coefficient is that when the time series anomaly score shows a high probability of anomaly but the residual anomaly score is low, there may be potential anomalies that have not been fully revealed. Therefore, through the gain coefficient, the anomaly signal under this condition can be amplified to ensure that these potential anomalies are not overlooked.

[0067] Further, based on the potential anomaly recognition gain coefficient, a high-confidence anomaly discrimination network compensation is performed to complete the anomaly fusion analysis. In this process, the system adjusts and compensates the anomaly score by applying the gain coefficient on the basis of the original anomaly score. This compensation mechanism ensures that in the case of potential anomalies, the system does not underestimate the severity of the anomalies, thus making a more accurate anomaly judgment. Through this compensation mechanism, more concealed anomalies can be captured in the complex and changeable charging pile operating environment, improving the sensitivity and accuracy of detection. Finally, this series of operations helps to improve the anomaly fusion analysis, ensuring that the charging pile can be effectively monitored and responded to in case of possible anomalies, and guaranteeing the safe and stable operation of the system.

[0068] Further, the application embodiment further includes step P80, and step P80 further includes:

[0069] P81: Configure a buffer interval, where the buffer interval includes a residual buffer interval and a time series buffer interval; P82: Perform discriminative proximity recognition of the high-confidence anomaly discrimination network through the buffer interval to generate a proximity recognition result; P83: If the high-confidence anomaly discrimination network does not identify an anomaly result and the proximity recognition result is a trigger result, generate a two-stage confirmation instruction; P84: Re-evaluate the anomaly status of the charging pile according to the two-stage confirmation instruction.

[0070] Among them, for re-evaluating the anomaly status of the charging pile according to the two-stage confirmation instruction, the evaluation formula is as follows:

[0071] Among them, S'(t) is the re-evaluated anomaly score, R(t - 1) is the residual anomaly score at the previous time node, R(t - 2) is the residual anomaly score at the two previous time nodes, T(t - 1) is the time series anomaly score at the previous time node, and T(t - 2) is the time series anomaly score at the two previous time nodes.

[0072] Specifically, to improve the reliability and accuracy of anomaly detection, the system introduces multiple operation steps, including configuring a buffer interval, performing proximity recognition, triggering a two-stage confirmation instruction, and re-evaluating the anomaly status of the charging pile.

[0073] First, configure a buffer interval, where the buffer interval includes a residual buffer interval and a time series buffer interval. The buffer interval provides a critical range for anomaly recognition, allowing fluctuations within a certain range without immediately judging as an anomaly, thereby reducing false alarms. The residual buffer interval and the time series buffer interval are respectively applied to the residual anomaly score and the time series anomaly score, providing a reference for the discrimination of the high-confidence anomaly discrimination network.

[0074] Furthermore, discriminative proximity recognition of the high-confidence anomaly discrimination network is performed through the buffer zone to generate a proximity recognition result. Discriminative proximity recognition means that in the high-confidence anomaly discrimination network, when a score is detected to be close but not exceeding the buffer zone, a proximity recognition result is generated to indicate that an anomaly may be about to occur. This result provides a basis for the next two-stage confirmation.

[0075] In the case where the high-confidence anomaly discrimination network does not identify an anomaly result, if the proximity recognition result shows a trigger result, the system will generate a two-stage confirmation instruction. The trigger result indicates that although no clear anomaly has been identified yet, the score detected by the system is close to the anomaly threshold and an anomaly may occur at a future time point, so further confirmation is required.

[0076] Therefore, according to the two-stage confirmation instruction, the anomaly status of the charging pile is re-evaluated. At this time, the system will use a new formula to re-evaluate the anomaly status, and the formula is as follows:

[0077] Among them, S'(t) is the re-evaluated anomaly score, which is calculated after considering the average score of the past three time nodes. By averaging the scores of multiple time nodes, the anomaly trend can be evaluated more smoothly, reducing misjudgments caused by fluctuations at a single moment. R(t - 1) is the residual anomaly score of the previous time node, and R(t - 2) is the residual anomaly score of the two previous time nodes. The average of these scores represents the performance trend of the residual anomaly over a long period. T(t - 1) is the time series anomaly score of the previous time node, and T(t - 2) is the time series anomaly score of the two previous time nodes. Similarly, the average of these scores reflects the fluctuations of the time series anomaly over multiple time periods.

[0078] Through this re-evaluation method, the system can more accurately judge the anomaly status of the charging pile. Especially when the original score result is close to the threshold but does not trigger an anomaly, the two-stage confirmation can provide a more reliable basis for anomaly determination, ensuring timely and accurate response to the anomaly status.

[0079] Furthermore, the application embodiment further includes step P90: establishing an anomaly warning level, performing trigger analysis of the anomaly warning level according to the anomaly status detection result, configuring a warning response plan according to the trigger analysis result, and performing anomaly management of the charging pile based on the warning response plan.

[0080] It should be understood that an anomaly warning level is further established to better handle possible anomalies of the charging pile. The anomaly warning level is a set of multiple alarm levels based on the anomaly status detection result, and these levels reflect the severity and urgency of the anomaly situation.

[0081] First, trigger analysis of the abnormal warning level is performed based on the abnormal state detection results generated in the previous steps. Trigger analysis refers to the system's evaluation of the detection results to determine whether the anomaly has reached the threshold for triggering a warning and to determine the warning response level. The process of trigger analysis usually involves a comprehensive consideration of anomaly scoring, trend analysis, historical data, and current environmental parameters to ensure the accuracy of the warning.

[0082] According to the results of the trigger analysis, the corresponding warning response plan is configured. The warning response plan refers to the specific countermeasures taken by the system after identifying different warning levels. These measures may include automatically adjusting the operating parameters of the charging pile, notifying maintenance personnel to conduct on-site inspections, and even shutting down the charging pile in extreme cases to prevent safety accidents. Different warning levels correspond to different response plans to ensure that the system can respond quickly and appropriately in different degrees of abnormal situations.

[0083] Finally, based on the configured warning response plan, abnormal management of the charging pile is carried out. Abnormal management covers the entire process from warning trigger to final handling, ensuring that anomalies can be identified and disposed of in a timely manner, reducing the impact of anomalies on the operation of the charging pile. Through this method, the system can not only detect anomalies but also automatically take corresponding measures according to their severity, thus ensuring the safety and stability of the charging pile.

[0084] In summary, the embodiments of the present application at least have the following technical effects:

[0085] The present application detects vehicle state anomalies by reading real-time collaborative information, establishes a first anomaly recognition result, performs grid anomaly recognition of long and short sequences, establishes a second anomaly recognition result, performs residual anomaly detection through the joint impact analysis of the first and second anomaly recognition results, generates a third anomaly recognition result, and performs time series change analysis of the charging pile operation data information, generates a fourth anomaly recognition result, and fuses it with the third anomaly recognition result to generate an abnormal state detection result.

[0086] It achieves the technical effect of combining vehicle, charging pile, and grid data to perform multi-level anomaly recognition and exclusion analysis, improving the accuracy and timeliness of abnormal state detection of charging facilities.

[0087] It should be noted that the above sequence of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. In addition, the specific embodiments of this specification have been described. Also, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-tasking and parallel processing are also possible or may be advantageous.

[0088] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

[0089] This specification and the drawings are merely illustrative of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A method for detecting abnormal state of charging facilities in a vehicle-pile-network collaborative manner, characterized in that: The method comprises: Reading real-time collaborative information, the real-time collaborative information includes vehicle status data information, charging pile operation data information, and power grid monitoring data information, and the real-time collaborative information is information aligned by timestamp; Performing data analysis on the real-time collaborative information, configuring an abnormality identification channel that matches the vehicle according to the analysis result, performing abnormality detection on the vehicle based on the abnormality identification channel, and establishing a first abnormality identification result; Based on the power grid monitoring data information, call the power grid traceback data, perform long and short sequence power grid anomaly identification based on the power grid traceback data and the power grid monitoring data information, and establish a second anomaly identification result; Establishing a joint anomaly elimination rule, performing a joint impact analysis of the first anomaly recognition result and the second anomaly recognition result based on the joint anomaly elimination rule, and establishing compensation residual data; Perform residual abnormality detection of the charging pile using the compensated residual data and the charging pile operation data information to generate a third abnormality recognition result; configuring a sliding window, and performing a time series change analysis of the charging pile operation data information based on the sliding window to generate a fourth abnormality recognition result; An abnormal fusion analysis is performed on the third abnormal recognition result and the fourth abnormal recognition result to generate an abnormal state detection result of the charging pile.

2. The method for detecting abnormal state of charging facilities in a vehicle-pile-network collaborative manner as claimed in claim 1, characterized in that: The configuration sliding window also includes: Using the charging pile operation data information as input data, and inputting it into a data trend fluctuation identification model; Obtain the fluctuation recognition result of the data trend fluctuation recognition model, use the fluctuation recognition result as matching data, perform data matching of the window mapping sequence, and generate the size constraint of the sliding window; The sliding window is configured according to the size constraint of the sliding window.

3. The method for detecting abnormal state of charging facilities in a vehicle-pile-network collaborative manner as claimed in claim 2, characterized in that: The time series change analysis of the charging pile operation data information based on the sliding window also includes: The time series change analysis of charging pile operation data information is carried out through the formula as follows: Where SW(t) represents the sliding window score of the operating parameters calculated at time point t, W is the size of the sliding window, t is the time point, i represents the current loop index, and F c (i) Characteristic value representing the charging pile operation data at time point i, F c (i-1) represents the characteristic value of the charging pile operation data at time point i-1.

4. The method for detecting abnormal state of charging facilities in a vehicle-pile-network coordinated manner as claimed in claim 1, characterized in that: The performing abnormal fusion analysis on the third abnormal recognition result and the fourth abnormal recognition result to generate an abnormal state detection result of the charging pile also includes: Configure a high-confidence anomaly discrimination network. The high-confidence anomaly discrimination network is as follows: Among them, S(t) represents the comprehensive anomaly score, R(t) is the anomaly score of the third anomaly recognition result, T(t) is the anomaly score of the fourth anomaly recognition result, θ R is the threshold of residual abnormality score, θ T is the threshold for time series anomaly scoring, w R Characterizes the weight of residual abnormal evaluation, w T Characterize the weight of time series anomaly evaluation, Characterizes the indicator function, which is used for conditional judgment. When the condition in the brackets is true, otherwise Complete anomaly fusion analysis based on high-confidence anomaly discrimination network.

5. The method for detecting abnormal state of charging facilities in a vehicle-pile-network collaborative manner as claimed in claim 4, characterized in that: The abnormal fusion analysis is completed according to the high-confidence abnormality discrimination network, and further includes: Configure the potential anomaly recognition gain coefficient, the activation condition of the potential anomaly recognition gain coefficient is T(t)≥θ T , and R(t)<θ R ; A high-confidence anomaly discrimination network compensation is performed according to the potential anomaly recognition gain coefficient to complete the anomaly fusion analysis.

6. The method for detecting abnormal state of charging facilities in a vehicle-pile-network coordinated manner as claimed in claim 4, characterized in that: The method further comprises: Configuring buffer spaces, wherein the buffer spaces include residual buffer spaces and time series buffer spaces; The proximity recognition is performed by using a high-confidence abnormality recognition network between the buffers to generate a proximity recognition result; If the high-confidence anomaly discrimination network does not identify an abnormal result, and the close identification result is a trigger result, a second-stage confirmation instruction is generated; The abnormal state of the charging pile is re-evaluated according to the second-stage confirmation instruction.

7. The method for detecting abnormal state of charging facilities in a vehicle-pile-network coordinated manner as claimed in claim 6, characterized in that: The abnormal state of the charging pile is re-evaluated according to the second-stage confirmation instruction, and the evaluation formula is as follows: Among them, S'(t) is the re-evaluated anomaly score, R(t-1) is the residual anomaly score of the previous time node, R(t-2) is the residual anomaly score of the previous two time nodes, T(t-1) is the time series anomaly score of the previous time node, and T(t-2) is the time series anomaly score of the previous two time nodes.

8. The method for detecting abnormal state of charging facilities in a vehicle-pile-network collaborative manner as claimed in claim 1, characterized in that: The method further comprises: Establish an abnormal warning level, perform a trigger analysis of the abnormal warning level according to the abnormal state detection result, configure a warning response plan according to the trigger analysis result, and perform abnormal management of the charging pile based on the warning response plan.

Images