A supercharging host power line protection method and system

Abstract

The present invention relates to the field of data processing, and specifically provides a supercharging host power line protection method and system, which obtains a high-frequency power waveform data stream for abnormal identification; determines a reference time window data segment from each time window data segment therein; for each time window data segment in the data stream, determines the waveform feature commonality measurement between the time window data segment and the reference time window data segment, and obtains the waveform commonality measurement index of the time window data segment; determines the waveform distortion mutation degree between adjacent time windows in the high-frequency power waveform data stream according to the waveform commonality measurement index corresponding to each time window data segment in the high-frequency power waveform data stream; determines the fault time window data segment from each time window data segment of the high-frequency power waveform data stream based on the waveform distortion mutation degree between adjacent time windows; thereby completing the solid-state circuit breaker tripping action of the supercharging host. The present invention can improve the intervention speed and accuracy of line protection.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a supercharging host power line protection method and system. Background Art

[0002] With the rapid development of the electric vehicle industry, superchargers are increasingly being used in charging facilities. Superchargers can achieve fast charging, greatly shortening the charging time of electric vehicles and improving user convenience. However, during the operation of superchargers, their power lines face many safety challenges.

[0003] When the output power of the charging pile changes in a step, the high-frequency power waveform of the target power line will undergo complex changes. This step change may be caused by a variety of factors, such as the start and stop of the charging equipment, or a sudden load change in the power system. Traditional power line protection methods have certain limitations when dealing with such high-frequency, rapidly changing power waveforms.

[0004] Existing protection methods often have difficulty accurately identifying abnormal features in power waveforms. During power step changes, waveform distortion and mutation may occur instantly, and traditional methods may not be able to capture these subtle but critical changes in time. In addition, traditional protection methods are not accurate enough in determining the time and location of faults, and cannot quickly locate the fault time window data segment, making it impossible to take effective protection measures in time when a fault occurs, which may cause damage to the supercharger host and power lines, and even cause safety accidents. Summary of the invention

[0005] In view of this, the present invention provides a supercharging host power line protection method and system. The solution of the present invention is implemented as follows:

[0006] According to one aspect of the present invention, a supercharging host power line protection method is provided, the method comprising: obtaining a high-frequency power waveform data stream to be used for abnormal identification; the high-frequency power waveform data stream is collected when the output power of the charging pile of the target power line undergoes a step change; determining a reference time window data segment in each time window data segment of the high-frequency power waveform data stream; for each time window data segment in the high-frequency power waveform data stream, determining a waveform feature commonality measure between the time window data segment and the reference time window data segment, and obtaining a waveform commonality measure index corresponding to the time window data segment; based on each time window data segment in the high-frequency power waveform data stream, determining a waveform feature commonality measure between the time window data segment and the reference time window data segment, and obtaining a waveform commonality measure index corresponding to the time window data segment; The waveform commonality measurement indicators corresponding to the data segments are used to determine the degree of waveform distortion mutation between adjacent time windows in the high-frequency power waveform data stream; based on the degree of waveform distortion mutation between adjacent time windows, the fault time window data segment is determined in each of the time window data segments of the high-frequency power waveform data stream; the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment of the fault time window data segment is the largest; when it is detected that the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment of the fault time window data segment is greater than the preset protection threshold, the solid-state circuit breaker of the supercharging host is triggered to perform a tripping action.

[0007] According to another aspect of the present invention, a computer system is provided, comprising: a processor; and a memory, wherein the memory stores a computer readable code, and when the computer readable code is executed by the processor, the processor executes the method described above.

[0008] Beneficial effects: The supercharging host power line protection method and system provided by the present invention are based on obtaining a high-frequency power waveform data stream for abnormal identification, which is collected when the output power of the charging pile of the target power line changes in a step, and determining a reference time window data segment in each time window data segment of the high-frequency power waveform data stream. For each time window data segment in the high-frequency power waveform data stream, the waveform feature commonality measurement between the time window data segment and the reference time window data segment is determined to obtain a waveform commonality measurement index corresponding to the time window data segment, and the waveform distortion mutation degree between adjacent time windows in the high-frequency power waveform data stream is determined based on the waveform commonality measurement index corresponding to each time window data segment in the high-frequency power waveform data stream. Then, based on the waveform distortion mutation degree between adjacent time windows, the fault time window data segment is determined from each time window data segment of the high-frequency power waveform data stream, and the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment of the fault time window data segment is the largest. Compared with the manual determination of time window data segments in the prior art, the present invention can automatically determine the time window data segments in the high-frequency power waveform data stream, and use this to open the solid-state circuit breaker of the supercharging host, thereby improving the intervention speed and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 It is a schematic diagram of the architecture of the application environment provided by the present invention;

[0010] Figure 2 It is a flow chart of a supercharging host power line protection method provided by the present invention;

[0011] Figure 3 It is a structural diagram of a computer system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0012] In order to facilitate a clearer understanding of the present invention, the application environment of the supercharging host power line protection method of the present invention is first introduced. Figure 1 As shown, the application environment may include a computer system 10 and a terminal cluster, and the terminal cluster may include one or more terminals, and the number of terminals is not limited here. Figure 1 As shown, the terminal cluster may specifically include terminal 1, terminal 2, ..., terminal n; it can be understood that terminal 1, terminal 2, terminal 3, ..., terminal n can all be connected to the computer system 10 through a network, so that each terminal can exchange data with the computer system 10 through the network connection.

[0013] It is understandable that the computer system 10 may refer to a hardware system for executing the method of the present invention, for example, it may be a server, a personal computer, a laptop, a virtual machine, or other device with computing capabilities. The terminal may specifically refer to a device for acquiring power data in a power system, such as a current sensor, a voltage sensor, and the like. Each terminal and the computer system 10 may be directly or indirectly connected via wired or wireless communication, and the number of terminals and computer systems may be one or at least two, and the present invention is not limited thereto.

[0014] See also Figure 2 , is a flow chart of a supercharger host power line protection method provided by an embodiment of the present invention. Figure 2 As shown, this method can be Figure 1 The computer system in the embodiment is used for execution, wherein the supercharging host power line protection method may include the following steps:

[0015] Step S100: obtaining a high-frequency power waveform data stream to be used for abnormality identification; the high-frequency power waveform data stream is collected when a step change occurs in the output power of a charging pile of a target power line.

[0016] In step S100, the high-frequency power waveform data stream refers to a continuous data sequence collected at a high sampling frequency in the power system, which reflects the changes in electrical quantities such as voltage and current in the power line over time. It can present the dynamic characteristics of the power waveform in detail, which is of great significance for detecting abnormal conditions in the power system. The target power line refers to the power line that needs to be monitored and protected, such as the line connecting the supercharging host and the charging pile. The output power of the charging pile undergoes a step change, which means that the output power of the charging pile undergoes a sudden and substantial change in a short period of time, such as the charging pile suddenly switching from a low-power output state to a high-power output state, or rapidly reducing from a high-power output state to a low-power output state.

[0017] When the computer system acquires the high-frequency power waveform data stream for abnormal identification, it can install voltage sensors and current sensors on the target power line. These sensors can measure the voltage and current values ​​in the line in real time. The data acquisition device is connected to the sensor, collects the voltage and current values ​​measured by the sensor at a high-frequency sampling rate, and converts the collected data into digital signals to form a high-frequency power waveform data stream. For example, on the power line of an electric vehicle charging pile, high-precision voltage sensors and current sensors are installed, and the data acquisition device collects the voltage and current at a sampling rate of 10,000 times per second, so that a detailed high-frequency power waveform data stream can be obtained. Or it can be realized through smart meters and communication networks. Smart meters have data acquisition and communication functions. They actually record various electrical parameters in the power line in real time and transmit these data to the computer system through the communication network. After receiving the data, the computer system processes and analyzes it to extract the high-frequency power waveform data stream.

[0018] In practical applications, in order to ensure the accuracy and reliability of the acquired high-frequency power waveform data stream, the computer system can preprocess the collected data. Preprocessing includes operations such as data filtering, data calibration and data normalization. Data filtering can remove noise and interference signals in the data and improve the quality of the data, such as mean filtering, median filtering and wavelet filtering. For example, using the mean filtering method, for a set of collected voltage data, the average value is calculated, and each data point is compared with the average value, and the data points with large deviations are removed to obtain smoother voltage waveform data. Data calibration is to ensure that the collected data is consistent with the actual electrical parameter values. Since there may be certain errors in sensors and data acquisition equipment, the collected data needs to be calibrated. For example, by comparing with standard voltage sources and current sources, the measured values ​​of voltage sensors and current sensors are calibrated to make the collected data more accurate. Data normalization is to convert the collected data into a unified numerical range for subsequent analysis and processing. Feasible data normalization methods include linear normalization and Z-score normalization. Through data normalization, the dimensional differences between different data can be eliminated and the accuracy of data analysis can be improved.

[0019] Step S200: determining a reference time window data segment in each time window data segment of the high-frequency power waveform data stream.

[0020] In step S200, the high-frequency power waveform data stream is divided into a plurality of time window data segments, each of which represents the power waveform data collected within a specific time period. The reference time window data segment is a data segment selected as a reference standard in the entire high-frequency power waveform data stream, and subsequent analysis will be based on the reference data segment and other time window data segments for comparison. When determining the reference time window data segment, different strategies will be adopted according to different types of fault time window data segments. The fault time window data segment may include a fault trigger time window data segment, a start time window data segment, and a termination time window data segment. When the fault time window data segment is a fault trigger time window data segment, the computer system comprehensively considers the power state before the step change and the characteristics of the trigger edge to determine the reference time window data segment. Before the step change, the power system is usually in a relatively stable state, but there may be some subtle changes close to the critical value of the fault trigger. The computer system screens out the time window data segment at the trigger edge by analyzing the electrical parameters of each time window data segment in the high-frequency power waveform data stream, such as the change trend and fluctuation of voltage and current. Then, the key monitoring data segments about the fault-sensitive areas are further selected from these data segments as the reference time window data segments. For example, in the power line of an electric vehicle charging pile, when the output power of the charging pile is about to undergo a step change, by monitoring the slight fluctuations in voltage and the instantaneous changes in current, the time window data segment close to the fault triggering critical value before the step change is identified. For certain specific fault types, such as short-circuit faults, the fault-sensitive areas may be areas where the voltage drops sharply or the current increases suddenly. The computer system focuses on the data in these areas and uses them as the reference time window data segments.

[0021] When the fault time window data segment is a startup time window data segment, the computer system searches for a time window data segment before the step change and at the edge of the step in the high-frequency power waveform data stream, and uses it as the reference time window data segment. Before the step start moment, the state of the power system may have changed in some ways, and these changes indicate an upcoming step. The computer system can determine the time window data segment at the edge of the step by analyzing the rate of change and trend of the power parameters. For example, in the process of switching the charging pile from low-power output to high-power output, the computer system can monitor the rising trend of the current and the falling trend of the voltage. When these trends reach a certain threshold, the corresponding time window data segment is considered to be a data segment at the edge of the step, and it is used as the reference time window data segment.

[0022] When the fault time window data segment is the termination time window data segment, the computer system determines the time window data segment after the step change is terminated in the high-frequency power waveform data stream, and uses it as the reference time window data segment. After the step change is terminated, the power system may enter a new stable state, and the data at this time can be used as a reference standard for subsequent analysis. For example, when the charging pile completes the power step change and enters a stable high-power output state, the computer system selects the corresponding time window data segment at this time as the reference time window data segment. If the high-frequency power waveform data stream stops collecting after the step change ends, then the step termination time window data segment may include the last time window data segment in the high-frequency power waveform data stream, and the computer system can directly use the last time window data segment as the reference time window data segment.

[0023] In order to accurately determine the reference time window data segment, the computer system can be implemented based on feature extraction and pattern recognition technology. Specifically, feature extraction can be performed on each time window data segment in the high-frequency power waveform data stream, such as extracting the amplitude, frequency, phase and other features of voltage and current. Then, these features are analyzed and classified through pattern recognition algorithms, such as cluster analysis, neural networks, etc., to determine the time window that meets the characteristics of the reference time window data segment. For example, a cluster analysis algorithm is used to divide the time window data segment into different categories, each category represents a different power state, and the computer system can select a category that is in a stable state and meets specific conditions as the reference time window data segment.

[0024] Alternatively, the reference time window data segment is determined by using a threshold-based judgment method. Specifically, some thresholds related to power parameters are set, such as voltage thresholds, current thresholds, etc. When the power parameters of the time window data segment meet these threshold conditions, the data segment may be selected as the reference time window data segment. For example, a threshold of a voltage fluctuation range is set. When the voltage fluctuation of a certain time window data segment is within the threshold range, the data segment can be considered as the reference time window data segment.

[0025] The computer system can also combine historical data and empirical knowledge to determine the reference time window data segment. By analyzing previous cases of similar power system failures or step changes, the characteristics and rules of the reference time window data segment are summarized. In practical applications, these historical data and empirical knowledge can be used to analyze and judge the current high-frequency power waveform data stream, so as to more accurately determine the reference time window data segment.

[0026] In addition, in order to improve the accuracy and reliability of determining the reference time window data segment, the computer system can also adopt a multi-dimensional analysis method. In addition to considering the changes in power parameters, it can also combine environmental factors, equipment operating status and other information for comprehensive analysis. For example, in a high temperature environment, the performance of power equipment may be affected, resulting in changes in power parameters. The computer system can include ambient temperature information in the analysis scope to more comprehensively evaluate the status of the power system, thereby determining a more appropriate reference time window data segment.

[0027] Step S300: for each time window data segment in the high-frequency power waveform data stream, determine the waveform feature commonality measurement between the time window data segment and the reference time window data segment, and obtain the waveform commonality measurement index corresponding to the time window data segment.

[0028] In step S300, the computer system determines the waveform feature commonality measurement between each time window data segment in the high-frequency power waveform data stream and the reference time window data segment, thereby obtaining the waveform commonality measurement index corresponding to the time window data segment. This step plays a key role in the subsequent analysis of power waveform changes and fault identification.

[0029] The waveform feature commonality metric is an indicator that measures the waveform similarity between two time window data segments, which reflects the characteristic consistency of the power waveform in different time periods. The waveform commonality metric is a quantitative representation of this similarity, which is convenient for computer systems to analyze and compare.

[0030] When determining the commonality measure of waveform features, multiple implementation methods can be used. One feasible method is based on correlation analysis. Correlation analysis can evaluate the degree of linear correlation between two time window data segments. The feasible correlation coefficient is the Pearson correlation coefficient. The calculation formula of the Pearson correlation coefficient is: ,in and are the electrical parameter values ​​at the corresponding moments in the two time window data segments, and are the means of the two time window data segments, and n is the number of data points. The value range of the correlation coefficient r is between -1 and 1. The closer r is to 1, the stronger the linear correlation between the two time window data segments and the higher the commonality of waveform features; the closer r is to -1, the weaker the linear correlation between the two time window data segments and the lower the commonality of waveform features; r is close to 0, indicating that there is almost no linear correlation between the two time window data segments.

[0031] Another feasible method is based on distance measurement. Distance measurement can measure the difference between two time window data segments. Viable distance measurement methods include Euclidean distance and Manhattan distance. Taking Euclidean distance as an example, its calculation formula is: ,in and are the electrical parameter values ​​at the corresponding moments in the two time window data segments, and n is the number of data points. The smaller the value of the Euclidean distance d, the smaller the difference between the two time window data segments, and the higher the commonality of the waveform features; the larger the value of d, the greater the difference between the two time window data segments, and the lower the commonality of the waveform features.

[0032] In addition to the above methods, the computer system can also use a feature matching-based method to determine the waveform feature commonality measure. This method first extracts features from the time window data segment, such as extracting the peak, valley, frequency, phase and other features of the waveform, and then compares the feature values ​​of the two time window data segments to calculate the similarity between the features. For example, by calculating the peak difference and frequency difference of the two time window data segments, their waveform feature commonality can be comprehensively evaluated. A calculation formula for feature similarity can be set, such as: , where S is the feature similarity, w j is the weight of the jth feature, f 1j and f 2j are the j-th eigenvalues ​​of the two time window data segments, f maxj is the maximum value of the jth feature, and m is the number of features. The closer the feature similarity S is to 1, the higher the commonality of the waveform features of the two time window data segments; the closer the value of S is to 0, the lower the commonality of the waveform features of the two time window data segments.

[0033] After determining the waveform feature commonality measure, the computer system converts it into a waveform commonality measure index. In order to make the waveform commonality measure index of different time window data segments comparable, it is usually necessary to convert it to a preset numerical range. A feasible method is to use a per-unit value method to divide the waveform feature commonality measure value by a reference value and normalize it to a specific interval. For example, waveform feature commonality measures such as correlation coefficients, distance measures, or feature similarity are divided by a pre-set reference value to obtain a waveform commonality measure index between 0 and 1. In this way, the computer system can easily compare and analyze the waveform commonality measure indexes of different time window data segments.

[0034] The computer system can also combine historical data and empirical knowledge to optimize the calculation of waveform feature commonality measurement and waveform commonality measurement indicators. By analyzing the waveform data of similar power system failures or normal operation in the past, the laws and characteristics of waveform feature commonality between different time window data segments are summarized. In practical applications, these historical data and empirical knowledge are referred to to adjust the calculation methods and parameters to improve the accuracy and reliability of waveform feature commonality measurement and waveform commonality measurement indicators.

[0035] Step S400: Determine the degree of waveform distortion mutation between adjacent time windows in the high-frequency power waveform data stream based on the waveform commonality measurement index corresponding to each time window data segment in the high-frequency power waveform data stream.

[0036] As mentioned above, the waveform commonality metric is a quantitative representation of the similarity between the waveform characteristics of each time window data segment in the high-frequency power waveform data stream and the reference time window data segment, while the waveform distortion mutation degree reflects the sudden change of the power waveform characteristics between adjacent time windows. By analyzing the waveform distortion mutation degree, the computer system can promptly detect possible faults or abnormal conditions in the power line.

[0037] When determining the degree of waveform distortion mutation between adjacent time windows, the computer system can use a variety of methods to implement it. One feasible method is based on difference calculation. Specifically, the difference between the waveform commonality metric indicators of adjacent time window data segments is calculated, and the size of the difference reflects the degree of waveform distortion mutation. For example, let the waveform commonality metric indicator of the i-th time window data segment be M i , the waveform commonality metric of the i+1th time window data segment is M i +1, then the degree of waveform distortion mutation between adjacent time windows can be expressed as the difference To express the difference The larger the absolute value of the difference is, the more severe the waveform distortion mutation between adjacent time windows is; The smaller the absolute value of is, the smoother the waveform changes between adjacent time windows are.

[0038] Another implementation method is based on ratio calculation. Specifically, the ratio relationship between the waveform commonality metrics of adjacent time window data segments can be calculated, and the change in ratio reflects the degree of waveform distortion mutation. For example, the degree of waveform distortion mutation between adjacent time windows can be expressed by the ratio R=M i+1 / M iWhen the ratio R deviates greatly from 1, it means that the waveform between adjacent time windows has changed significantly and there is a sudden change in waveform distortion; when the ratio R is close to 1, it means that the waveform between adjacent time windows has changed little. The threshold range of the ratio can be set according to the actual situation. When the ratio exceeds the threshold range, it is determined that there is a sudden change in waveform distortion between adjacent time windows.

[0039] In addition to difference calculation and ratio calculation, the computer system can also use a trend analysis-based method to determine the degree of waveform distortion mutation. By analyzing the trend of waveform commonality metrics over time, the computer system can determine whether the waveform between adjacent time windows has undergone a mutation. For example, the waveform commonality metrics are smoothed using the moving average method to obtain its change trend curve. Then, the slope change of the adjacent time window data segments on the trend curve is calculated. The sudden change in the slope indicates that the waveform has undergone a distortion mutation. Suppose the waveform commonality metrics sequence after moving average is , then the slope change between adjacent time windows can be expressed as To indicate that , t i is the time point of the i-th time window, is the waveform commonality metric at the time point of the i-th time window, It is a measure of the commonality of the waveform at the time point of the i+1th time window. When the threshold is exceeded, it indicates that the waveforms between adjacent time windows have undergone obvious distortion mutations.

[0040] The computer system can also combine historical data and empirical knowledge to optimize the judgment of the degree of waveform distortion mutation. By analyzing the waveform data of similar power system failures or normal operation in the past, the typical characteristics and threshold range of the degree of waveform distortion mutation are summarized. In practical applications, these historical data and empirical knowledge are used to evaluate the current degree of waveform distortion mutation to improve the accuracy and reliability of judgment.

[0041] After determining the degree of waveform distortion mutation between adjacent time windows, the computer system can judge the degree of mutation according to the set threshold. When the degree of waveform distortion mutation exceeds the preset threshold, it indicates that there may be a fault or abnormality in the power line, which requires further analysis and processing. The computer system can promptly feedback these abnormalities to the operator so that appropriate measures can be taken to deal with them and ensure the safe and stable operation of the power line.

[0042] Step S500: Determine the fault time window data segment in each time window data segment of the high-frequency power waveform data stream based on the degree of waveform distortion mutation between adjacent time windows; the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment of the fault time window data segment is the largest.

[0043] In step S500, the degree of waveform distortion mutation reflects the sudden change of power waveform characteristics between adjacent time windows, and the signs of fault occurrence are found by analyzing the degree of change. In the high-frequency power waveform data stream, the degree of waveform distortion mutation may be different in different time periods, and faults usually cause significant waveform mutations. Therefore, the adjacent time window with the largest waveform distortion mutation value is found from the numerous time window data segments, and the latter time window is the fault time window data segment.

[0044] When determining the fault time window data segment, the computer system can adopt the following implementation methods. One method is to compare the waveform distortion mutation degree of all adjacent time window data segments in the high-frequency power waveform data stream one by one, and record the waveform distortion mutation value between each adjacent time window. For example, suppose the waveform distortion mutation value between adjacent time windows i and i+1 is , the computer system calculates (n is the total number of time window data segments), and then find the maximum value among them and The next time window data segment in the corresponding adjacent time window is the fault time window data segment. In the power line of an electric vehicle charging pile, when the computer system analyzes the high-frequency power waveform data stream, it calculates and compares the voltage waveform distortion mutation degree of each pair of adjacent time window data segments. When it is found that the voltage waveform distortion mutation value between a certain adjacent time window reaches the maximum, the next time window in the adjacent time window is determined to be the fault time window data segment, which may mean that a short circuit or other fault occurred in the power line at this time point.

[0045] Another implementation method is based on threshold screening and sorting. Specifically, a threshold for the degree of waveform distortion mutation can be set, and the waveform distortion mutation values ​​of all adjacent time window data segments can be compared with the threshold to screen out adjacent time windows that exceed the threshold. Then, the waveform distortion mutation values ​​of these adjacent time windows that exceed the threshold are sorted to find the adjacent time window corresponding to the largest mutation value, and then determine the fault time window data segment. For example, the threshold is set to T, when Then, for all the adjacent time windows that meet the conditions Sort in descending order, the first The next time window in the corresponding adjacent time windows is the fault time window data segment. This method can reduce unnecessary comparisons and improve the efficiency of determining the fault time window data segment.

[0046] The computer system can also use dynamic monitoring and tracking methods to determine the fault time window data segment. In the process of collecting high-frequency power waveform data streams, the computer system calculates the degree of waveform distortion mutation between adjacent time windows in real time and monitors it dynamically. When it is found that the degree of waveform distortion mutation suddenly increases and exceeds a certain threshold, the mutation situation begins to be tracked, and the waveform distortion mutation value of the subsequent adjacent time windows is recorded. If the waveform distortion mutation value of a subsequent adjacent time window reaches the maximum, the next time window in the adjacent time window is determined to be the fault time window data segment. For example, during the operation of the power line, the computer system continuously monitors the changes in the voltage waveform. When it is detected that the degree of voltage waveform distortion mutation suddenly increases and exceeds the preset threshold, it begins to track the voltage waveform distortion mutation of the subsequent adjacent time windows, and finally determines the time window data segment where the fault occurs.

[0047] After the computer system determines the fault time window data segment, it can also further analyze the data segment to understand the specific type and severity of the fault. For example, by analyzing the voltage, current amplitude and phase characteristics of the fault time window data segment, it can be determined whether the fault is a short circuit fault, a circuit breaker fault or other types of faults. At the same time, according to the degree and duration of waveform distortion mutation, the severity of the fault is evaluated to provide more detailed information for subsequent fault processing.

[0048] Step S600: When it is detected that the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment of the fault time window data segment is greater than the preset protection threshold, the solid-state circuit breaker of the supercharging host is triggered to perform a tripping action.

[0049] As mentioned above, the fault time window data segment is the time window data segment determined by the computer system as a fault in the high-frequency power waveform data stream. The waveform distortion mutation value between it and the previous adjacent time window data segment reflects the degree of sudden change in the power waveform when the fault occurs. The preset protection threshold is a critical value pre-set according to the characteristics and safety requirements of the power system. When the waveform distortion mutation value exceeds the threshold, it indicates that a more serious fault may have occurred in the power line and protection measures need to be taken immediately.

[0050] A solid-state circuit breaker is an electrical device that can quickly cut off a circuit. It has the characteristics of fast response speed and high reliability, and plays an important role in the protection of the supercharging host power line.

[0051] In step S600, the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment is continuously monitored. For example, the voltage and current data in the power line are collected in real time by sensors installed on the power line, such as voltage sensors and current sensors, and these data are transmitted to the computer system for processing. The collected data are analyzed to calculate the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment. For example, the waveform distortion mutation value is calculated by the difference calculation, ratio calculation or trend analysis method mentioned above in the present invention.

[0052] After calculating the waveform distortion mutation value, the computer system compares it with the preset protection threshold. The setting of the preset protection threshold needs to comprehensively consider factors such as the rated parameters of the power system, the fault type and safety requirements. Different preset protection thresholds need to be set for different types of faults, such as short circuit faults, overload faults, etc. For example, for short circuit faults, since it will cause a sharp increase in current and a large waveform distortion mutation value, the preset protection threshold can be set relatively high; while for overload faults, the current increase is relatively small, and the preset protection threshold can be set relatively low. The computer system can use database management technology to store the preset protection thresholds for different types of faults in the database, and query and call them according to actual conditions.

[0053] When the computer system detects that the sudden change value of waveform distortion is greater than the preset protection threshold, it indicates that a serious fault may have occurred in the power line, and it is necessary to immediately trigger the solid-state circuit breaker of the supercharging host to perform the tripping action. In order to realize this triggering function, the computer system can be connected to the solid-state circuit breaker through a communication interface. The communication interface can use wired communication methods, such as Ethernet, RS-485, etc., or wireless communication methods, such as Bluetooth, Wi-Fi, etc. After detecting the fault, the computer system sends a tripping command to the solid-state circuit breaker through the communication interface. After receiving the tripping command, the solid-state circuit breaker quickly cuts off the circuit to prevent the fault from further expanding.

[0054] In summary, by real-time monitoring of the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment, and comparing it with the preset protection threshold, when the waveform distortion mutation value is greater than the preset protection threshold, the solid-state circuit breaker of the supercharger host is triggered through the communication interface to perform the tripping action, and at the same time, subsequent processing such as fault information recording, alarm notification, fault analysis and diagnosis, and solid-state circuit breaker status monitoring are carried out to ensure the safe and stable operation of the power system.

[0055] As an implementation mode, step S400, based on the waveform commonality measurement index corresponding to each time window data segment in the high-frequency power waveform data stream, determines the degree of waveform distortion mutation between adjacent time windows in the high-frequency power waveform data stream, including:

[0056] Step S410: modeling transient waveform characteristics according to waveform commonality metrics corresponding to each time window data segment in the high-frequency power waveform data stream, and obtaining waveform characteristic commonality metrics evolution trajectory corresponding to the high-frequency power waveform data stream;

[0057] Step S500, determining a fault time window data segment in each time window data segment of the high-frequency power waveform data stream based on the waveform distortion mutation degree between adjacent time windows, comprises:

[0058] Step S510: extracting differential features from the waveform feature commonality metric evolution trajectory to obtain a waveform commonality transient change rate curve corresponding to the high-frequency power waveform data stream;

[0059] Step S520: The time window data segment corresponding to the mutation extreme point of the waveform common transient change rate curve is used as the fault time window data segment determined in each time window data segment of the high-frequency power waveform data stream; wherein the mutation extreme point is used to indicate that the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment of the fault time window data segment is the largest.

[0060] Transient waveform feature modeling is a mathematical description and abstract representation of the time-varying law of waveform commonality metrics of each time window data segment in the high-frequency power waveform data stream. For example, curve fitting is performed to capture the characteristic changes of the power waveform in the transient process by establishing a suitable model. The waveform feature commonality metric evolution trajectory is the result of transient waveform feature modeling, which intuitively shows the evolution of waveform commonality metrics over time in the form of a curve. When performing transient waveform feature modeling, a variety of implementation methods can be used.

[0061] One feasible method is based on polynomial fitting. Specifically, a polynomial function is used to approximate the relationship between the waveform commonality measurement index and time. Suppose the sequence number of the time window data segment is t, and the corresponding waveform commonality measurement index is M(t), then an n-order polynomial function can be used To fit M(t), are the coefficients of the polynomial. The computer system can determine these coefficients by the least square method so that the sum of squares of the errors between the polynomial function P(t) and the actual waveform commonality metric M(t) is minimized.

[0062] Another implementation method is based on neural network modeling. Specifically, the serial number of the time window data segment can be used as input, and the corresponding waveform commonality measurement index can be used as output to build a neural network model. By training a large amount of historical data, the weights and thresholds of the neural network are adjusted so that the neural network can accurately predict the changes of the waveform commonality measurement index over time. For example, a three-layer feedforward neural network is constructed, with one neuron in the input layer representing the serial number of the time window data segment, several neurons in the hidden layer, and one neuron in the output layer representing the waveform commonality measurement index. After training, the neural network can output the corresponding waveform commonality measurement index prediction value according to the input time window data segment serial number, thereby obtaining the waveform feature commonality measurement evolution trajectory.

[0063] After obtaining the waveform feature commonality metric evolution trajectory, the computer system performs differential feature extraction to obtain the waveform commonality transient rate of change curve. Differential feature extraction is to perform a derivative operation on the waveform feature commonality metric evolution trajectory to obtain the curve of its derivative changing over time. The derivative represents the rate of change of the function, so the waveform commonality transient rate of change curve reflects the speed of change of the waveform commonality metric index over time. The method of numerical differentiation can be used to extract differential features. Numerical differentiation is to approximate the derivative of the function through discrete data points. Feasible numerical differentiation methods include forward differentiation, backward differentiation, and central differentiation. Taking central differentiation as an example, assuming that the waveform feature commonality metric evolution trajectory is M(t), the derivative at time point t can be approximately calculated using the central difference formula: , where h is the time step. The central difference formula can be applied to each time point on the waveform feature commonality metric evolution trajectory to obtain the corresponding derivative value, thereby constructing the waveform commonality transient rate of change curve. For example, on the waveform feature commonality metric evolution trajectory, a data point is taken every 0.1 second, and the derivative at each data point is calculated using the central difference formula to obtain the waveform commonality transient rate of change curve.

[0064] The sudden extreme value point of the waveform common transient rate of change curve refers to the point on the curve where the derivative suddenly changes dramatically. The time window data segment corresponding to these points usually indicates that a fault has occurred in the power line. The sudden extreme value point can be the maximum value point or minimum value point of the derivative, or the point where the absolute value of the derivative is the largest. When determining the sudden extreme value point, for example, an extreme value search algorithm is used to find the extreme value point by traversing the curve data points. Specifically, the waveform common transient rate of change curve can be compared point by point to find the maximum value point and the minimum value point of the derivative. At the same time, in order to avoid the influence of noise and interference, a threshold can be set. When the change of the derivative exceeds the threshold, it is regarded as a sudden extreme value point. For example, on the waveform common transient rate of change curve, when the absolute value of the derivative suddenly increases from 0.1 to 1 and exceeds the preset threshold value of 0.5, the point is a sudden extreme value point.

[0065] After determining the mutation extreme point, the corresponding time window data segment is used as the fault time window data segment. The determination of the fault time window data segment is of great significance for accurately identifying the time and location of the power line fault. By analyzing the waveform characteristics of the fault time window data segment, the type and severity of the fault can be further determined. For example, when the voltage waveform of the fault time window data segment drops sharply and the current waveform increases sharply, it may indicate that a short circuit fault has occurred in the power line; when the voltage waveform and current waveform of the fault time window data segment change slowly, it may indicate that an overload fault has occurred in the power line.

[0066] In order to improve the accuracy and reliability of transient waveform feature modeling and differential feature extraction, the computer system also needs to preprocess the collected high-frequency power waveform data stream. Preprocessing includes operations such as data filtering, data calibration and data normalization. Data filtering can remove noise and interference signals in the data and improve the quality of the data. Feasible data filtering methods include mean filtering, median filtering and wavelet filtering. Data calibration is to ensure that the collected data is consistent with the actual electrical parameter values. Since there may be certain errors in sensors and data acquisition equipment, the collected data needs to be calibrated. Data normalization is to convert the collected data into a unified numerical range for subsequent analysis and processing. Feasible data normalization methods include linear normalization and Z-score normalization.

[0067] When modeling transient waveform features and extracting differential features, the time synchronization of data also needs to be considered. Since there may be time deviations between various sensors and data acquisition devices in the power system, in order to ensure that the collected high-frequency power waveform data stream can accurately reflect the actual operation of the power line, the data needs to be synchronized. GPS clock synchronization technology can be used to provide a unified time reference for each sensor and data acquisition device through GPS satellite signals to ensure the consistency of the collected data in time.

[0068] As an implementation method, the fault time window data segment includes a fault trigger time window data segment, and the fault trigger time window data segment is a time window data segment triggered by a fault. Based on this, step S200 determines a reference time window data segment in each time window data segment of the high-frequency power waveform data stream, including:

[0069] Step S210: determining a pre-fault reference time window data segment in each time window data segment of the high-frequency power waveform data stream, wherein the pre-fault reference time window data segment is a time window data segment in the high-frequency power waveform data stream that is before the step change and at the trigger edge;

[0070] Step S220: taking the key monitoring data segment about the fault-sensitive area in the pre-failure reference time window data segment as the reference time window data segment.

[0071] When executing step S210, find the pre-fault reference time window data segment from the numerous time window data segments of the high-frequency power waveform data stream. The pre-fault reference time window data segment has two key features: one is the time period before the step change in the output power of the charging pile, and the other is the edge state of the fault trigger. Before the step change means that the power system state corresponding to the time window data segment is relatively stable, but is about to face a sudden change in power; being on the edge of the trigger means that certain parameters of the power system are close to the critical value of the fault, and a slight change may cause a fault.

[0072] There are many ways to implement the computer system to determine the pre-fault reference time window data segment. One feasible method is based on threshold judgment. The computer system can set corresponding threshold ranges for key parameters in the power system, such as voltage, current, power, etc. When the parameter value in a certain time window data segment is close to but has not exceeded these threshold ranges, and a step change is immediately followed, the time window data segment may be determined as a pre-fault reference time window data segment. For example, for voltage parameters, the voltage range for normal operation is set to 210V-230V, and the critical voltage for fault triggering is 240V. When the computer system detects that the average voltage in a certain time window data segment reaches 235V, and a step change in the output power of the charging pile will soon occur, then this time window data segment may be a pre-fault reference time window data segment.

[0073] Another implementation method is based on trend analysis. Specifically, the parameter change trend of each time window data segment in the high-frequency power waveform data stream can be analyzed. By calculating the change rate of the parameter, such as the voltage change rate (in is the change in voltage, is the time interval), current change rate Etc., to determine whether the state of the power system is close to the fault triggering edge. If the change rate of the parameter in a certain time window data segment continues to increase and is close to the preset change rate threshold, and is before a step change, then the time window data segment may be a pre-fault reference time window data segment. For example, when the current change rate continues to rise and reaches the preset current change rate threshold close to the fault, and no step change occurs at this time, then the corresponding time window data segment can be used as a candidate for the pre-fault reference time window data segment.

[0074] In addition, machine learning algorithms can also be combined to determine the pre-fault reference time window data segment. For example, a classification algorithm such as a support vector machine (SVM) is used. Multiple parameters of each time window data segment in the high-frequency power waveform data stream are used as input features, and whether it is a pre-fault reference time window data segment is used as a classification label. A classification model is obtained by training a large amount of historical data. When the parameters of a new time window data segment are input, the classification model can determine whether the data segment is a pre-fault reference time window data segment. For example, a large amount of operating data of the charging pile power line in the past period of time is collected, including parameters such as voltage, current, power, and the corresponding labels of whether it is a pre-fault state. These data are used to train the SVM model, and then the model is used to classify and judge the current time window data segment.

[0075] In practical applications, taking an electric vehicle charging pile as an example, the computer system continuously monitors the high-frequency power waveform data stream of the charging pile power line. When the charging pile is about to switch from low-power output to high-power output, the computer system finds through threshold judgment that in a certain time window data segment before the step change, the current value has approached but not reached the overcurrent protection threshold. At the same time, through trend analysis, it is found that the current change rate is continuously increasing, approaching the preset dangerous change rate threshold. Combined with the judgment results of the machine learning model, this time window data segment is determined to be the pre-fault reference time window data segment.

[0076] After determining the pre-fault reference time window data segment, enter step S220, and find the key monitoring data segment about the fault-sensitive area from the pre-fault reference time window data segment, and use it as the reference time window data segment. Fault-sensitive areas refer to those areas in the power system that are easily affected by faults or where parameter changes are more obvious when faults occur. The key monitoring data segment is the data portion related to the fault-sensitive area in the pre-fault reference time window data segment.

[0077] There are also many ways to implement the computer system to determine the key monitoring data segments of fault-sensitive areas. One way is based on fault type analysis. Different types of faults will cause significant changes in parameters in different areas of the power system. For example, a short-circuit fault usually causes a sharp increase in current and a sharp drop in voltage near the fault point; while a grounding fault may cause abnormal changes in the voltage and current of the grounding phase. According to the feasible fault types, it is possible to analyze which areas of the pre-fault reference time window data segment have parameters that best match the change characteristics related to these fault types, thereby determining the key monitoring data segments of the fault-sensitive areas. For example, for short-circuit faults, focus on the data in the areas where the current suddenly increases and the voltage suddenly decreases in the pre-fault reference time window data segment, and use this part of the data as the key monitoring data segment.

[0078] Another implementation method is based on historical fault data statistics. Specifically, relevant data from past power system faults can be collected and analyzed to find out which areas have the most frequent and significant parameter changes when faults occur, thereby determining fault-sensitive areas. Then, the time and parameter data corresponding to these fault-sensitive areas are extracted in the pre-fault reference time window data segment as key monitoring data segments. For example, through statistical analysis of fault data over the past year, it was found that the temperature and resistance changes in a specific cable joint area were most obvious when a short-circuit fault occurred. Then, in the pre-fault reference time window data segment, the computer system will focus on the temperature and resistance data corresponding to the cable joint area and use it as a key monitoring data segment.

[0079] In addition, feature extraction and matching methods can also be used to determine the key monitoring data segments. Feature extraction is performed on the pre-fault reference time window data segment, such as extracting the voltage, current amplitude, frequency, phase and other features. These features are then matched with the feature templates of known fault-sensitive areas to find the data with the highest matching degree as the key monitoring data segment. For example, it is known that when a certain type of fault occurs, the voltage amplitude will fluctuate significantly at a specific frequency. The voltage amplitude and frequency features are extracted from the pre-fault reference time window data segment, matched with the feature template of the fault, and the time window and data that meet the feature template are found as the key monitoring data segment.

[0080] As another implementation, the fault time window data segment includes a start time window data segment, the step change process includes a step start moment, and the start time window data segment is a time window data segment reaching the step start moment. Based on this, step S200 determines a reference time window data segment in each time window data segment of the high-frequency power waveform data stream, including:

[0081] Step S201: Determine a pre-fault reference time window data segment in each time window data segment of the high-frequency power waveform data stream, and use the pre-fault reference time window data segment as the reference time window data segment; wherein the pre-fault reference time window data segment is a time window data segment in the high-frequency power waveform data stream before the step change and at the edge of the step.

[0082] As mentioned above, a step change refers to a sudden and substantial change in the output power of a charging pile in a short period of time, and the pre-fault reference time window data segment is at the stage where this step change is about to occur but has not yet occurred, and the state of the power system is close to the critical state of the step.

[0083] There are many ways to implement the computer system to determine the pre-fault reference time window data segment. One feasible method is based on threshold judgment. The computer system can set corresponding threshold ranges for key parameters in the power system, such as voltage, current, power, etc. When the parameter value in a time window data segment is close to but has not exceeded these threshold ranges, and a step change is immediately followed, the time window data segment may be determined as a pre-fault reference time window data segment. For example, for current parameters, the current range for normal operation is set to 0A-50A, and the critical current for step change is 55A. When the computer system detects that the average current in a time window data segment reaches 52A, and a step change in the output power of the charging pile will soon occur, then this time window data segment may be a pre-fault reference time window data segment.

[0084] Another implementation method is based on trend analysis. The computer system can analyze the parameter change trend of each time window data segment in the high-frequency power waveform data stream. By calculating the rate of change of parameters, such as the voltage change rate, the current change rate, etc., it is determined whether the state of the power system is close to the step edge. If the rate of change of the parameters in a time window data segment continues to increase and is close to the preset rate of change threshold, and is before the step change, then the time window data segment may be a pre-fault reference time window data segment. For example, when the current change rate continues to rise and reaches the preset current change rate threshold close to the step, and no step change has occurred at this time, then the corresponding time window data segment can be used as a candidate for the pre-fault reference time window data segment.

[0085] The computer system can also combine machine learning algorithms to determine the pre-fault reference time window data segment. For example, a classification algorithm is used, such as a decision tree algorithm. Multiple parameters of each time window data segment in the high-frequency power waveform data stream are used as input features, and whether it is a pre-fault reference time window data segment is used as a classification label. A classification model is obtained by training a large amount of historical data. When the parameters of a new time window data segment are input, the classification model can determine whether the data segment is a pre-fault reference time window data segment. For example, a large amount of operating data of the charging pile power line in the past period of time is collected, including parameters such as voltage, current, power, and the corresponding labels of whether it is a pre-fault state. These data are used to train a decision tree model, and then the model is used to classify and judge the current time window data segment.

[0086] After the pre-fault reference time window data segment is determined, it is used as the reference time window data segment. The reference time window data segment has important reference value in subsequent analysis. In the subsequent steps, other time window data segments in the high-frequency power waveform data stream are compared with this reference time window data segment to determine the commonality measurement of waveform characteristics and the degree of waveform distortion mutation. For example, by calculating the correlation and per-unit value between other time window data segments and the reference time window data segment, the waveform commonality measurement index is obtained, and then the waveform distortion mutation between adjacent time windows is analyzed to identify possible faults.

[0087] As another implementation, the fault time window data segment includes a termination time window data segment, the step change process includes a step termination moment, and the termination time window data segment is a time window data segment reaching the step termination moment. Based on this, step S200 determines a reference time window data segment in each time window data segment of the high-frequency power waveform data stream, including:

[0088] Step S200A: Determine the step termination time window data segment in each time window data segment of the high-frequency power waveform data stream, and use the step termination time window data segment as the reference time window data segment; wherein the step termination time window data segment is the time window data segment in the high-frequency power waveform data stream after the step change is terminated.

[0089] Since the power system will enter a relatively stable state after the step change is terminated, the data of this stage is used as the benchmark time window data segment, which can provide a reliable reference standard for subsequent comparative analysis with other time window data segments. The computer system can use a variety of implementation methods to determine the step termination time window data segment. One feasible method is based on parameter stability judgment. The key parameters in the high-frequency power waveform data stream, such as voltage, current, power, etc., can be monitored and analyzed in real time. During the step change process, these parameters will fluctuate significantly, and when the step change is terminated, the parameters will gradually stabilize. The parameter fluctuation threshold can be set. When the parameter fluctuation amplitude in a certain time window data segment is less than the threshold, it is considered that the power system has entered a stable state, and the time window data segment is the step termination time window data segment. For example, for voltage parameters, the fluctuation threshold is set to ±0.5V. When the computer system detects that the voltage fluctuation in a certain time window data segment is always within the range of ±0.4V, and the time window is after the step change, then the time window data segment can be determined as the step termination time window data segment.

[0090] Another implementation method is based on time stamps and event triggering. In actual power systems, the start and end of step changes are often accompanied by specific events or operations. The step end time window data segment can be determined by recording the time stamps of these events. For example, when the power regulation system of the charging pile performs a step change operation, it sends start and end signals to the computer system. The computer system determines the step end time window data segment based on the timestamps of these signals and the sampling time of the high-frequency power waveform data stream.

[0091] The computer system can also use pattern recognition methods to determine the step termination time window data segment. By learning and analyzing a large amount of historical step change data, the computer system can establish a pattern model for step changes. When processing the current high-frequency power waveform data stream, the computer system matches the data with the pattern model to find the time window data segment that meets the step termination characteristics. For example, using the clustering algorithm in machine learning, the data of the step termination stage in the historical data is clustered and analyzed to obtain the characteristic pattern of step termination. When a time window data segment in the current high-frequency power waveform data stream has a high degree of match with the characteristic pattern, it is determined as the step termination time window data segment.

[0092] After the step termination time window data segment is determined, it is used as the reference time window data segment. The reference time window data segment plays an important reference role in the subsequent power waveform analysis. In the subsequent steps, the computer system compares other time window data segments in the high-frequency power waveform data stream with this reference time window data segment to determine information such as the waveform feature commonality measurement and waveform distortion mutation degree between them. For example, by calculating the correlation and per-unit value between other time window data segments and the reference time window data segment, the waveform commonality measurement index is obtained, and then the waveform distortion mutation between adjacent time windows is analyzed to identify possible faults.

[0093] As an implementation method, the high-frequency power waveform data stream is obtained by stopping the collection after the step change ends, and the step termination time window data segment includes the last time window data segment in the high-frequency power waveform data stream. Based on this, step S200A, determining the step termination time window data segment in each time window data segment of the high-frequency power waveform data stream, and using the step termination time window data segment as the reference time window data segment, includes:

[0094] Step S200A1: Determine the last time window data segment in each time window data segment of the high-frequency power waveform data stream, and use the last time window data segment as a reference time window data segment.

[0095] The high-frequency power waveform data stream consists of a series of time window data segments arranged in chronological order, and each time window data segment records the relevant electrical parameters of the power line in a specific time period, such as voltage, current, etc. The determination of the last time window data segment can be achieved through counting and indexing operations. When collecting the high-frequency power waveform data stream, a unique serial number is assigned to each time window data segment, and the serial number increases as the collection process proceeds. When the collection stops, the time window data segment with the largest serial number is the last time window data segment. The last time window data segment is used as the reference time window data segment because after the step change ends, the power system usually enters a relatively stable state, and the last time window data segment can best reflect the power waveform characteristics in this stable state. In the subsequent analysis, the other time window data segments in the high-frequency power waveform data stream are compared with this reference time window data segment to determine the changes in the power waveform in different time periods, and then identify possible faults. For example, by calculating the waveform feature commonality measurement between other time window data segments and the reference time window data segment, the waveform commonality measurement index is obtained. If the waveform commonality measurement index of a certain time window data segment is significantly different from that of the reference time window data segment, it may mean that there is an abnormality in the power line during the time period corresponding to the time window data segment.

[0096] As an implementation mode, step S100, obtaining a high-frequency power waveform data stream to be subjected to abnormality identification, includes:

[0097] Step S110: Acquire an initial high-frequency power waveform data stream; the initial high-frequency power waveform data stream is collected when the output power of the charging pile of the target power line undergoes a step change;

[0098] Step S120: for each time window data segment in the initial high-frequency power waveform data stream, based on the instantaneous gradient of the electrical quantity between the time window data segment and the previous adjacent time window data segment of the time window data segment, obtain the time window waveform difference corresponding to the time window data segment;

[0099] Step S130: According to the time window waveform differences corresponding to each time window data segment in the initial high-frequency power waveform data stream, some time window data segments are selected from each time window data segment of the initial high-frequency power waveform data stream to obtain a high-frequency power waveform data stream for abnormality identification; wherein the time window waveform differences corresponding to the selected time window data segments are greater than the time window waveform differences corresponding to the remaining time window data segments; the remaining time window data segments are the time window data segments in the initial high-frequency power waveform data stream except the selected time window data segments.

[0100] When executing step S110, the computer system obtains the initial high-frequency power waveform data stream. This data stream reflects the change in the electrical state of the target power line when the output power of the charging pile changes stepwise. In order to obtain this data stream, the computer system usually uses sensors installed on the target power line, such as voltage sensors and current sensors. These sensors can monitor the voltage and current values ​​in the power line in real time, and convert the monitored analog signals into digital signals and transmit them to the computer system. The computer system collects these digital signals at a higher sampling frequency to form an initial high-frequency power waveform data stream. For example, in the power line of an electric vehicle charging pile, high-precision voltage sensors and current sensors are installed. The computer system collects voltage and current signals at a sampling frequency of 10,000 times per second. When the output power of the charging pile changes stepwise, the initial high-frequency power waveform data stream recording this change process is obtained.

[0101] In step S120, the time window waveform difference corresponding to each time window data segment is calculated. The time window data segment is obtained by dividing the initial high-frequency power waveform data stream according to a certain time interval, and each time window data segment contains the power waveform data within the time period. The instantaneous gradient of an electrical quantity represents the rate of change of the electrical quantity in a very short time. By calculating the instantaneous gradient of the electrical quantity between adjacent time window data segments, the degree of waveform difference between two time window data segments can be measured. Feasible electrical quantities include voltage and current. Taking voltage as an example, the instantaneous gradient of the electrical quantity can be represented by the voltage change rate. Suppose the voltage average value of the i-th time window data segment is , the average voltage of the i-1th time window data segment is , the duration of the time window is , then the instantaneous voltage gradient between the i-th time window data segment and the i-1-th time window data segment is According to the instantaneous gradients of different electrical quantities, the waveform difference of the time window is obtained comprehensively. For example, the weighted average method can be used, and the weight of the instantaneous gradient of voltage is set to , the weight of the instantaneous current gradient is , the instantaneous voltage gradient is , the instantaneous current gradient is , then the time window waveform difference corresponding to the i-th time window data segment is .

[0102] In step S130, the computer system selects some time window data segments according to the time window waveform difference of each time window data segment to obtain a high-frequency power waveform data stream for abnormal identification. Time window data segments with larger time window waveform differences often contain more power waveform change information, which may be related to abnormal conditions in the power line. The computer system can use a variety of methods to select data segments. One way is to set a time window waveform difference threshold T, and select the time window data segments with a time window waveform difference greater than the threshold T. For example, setting the threshold T = 5, the computer system compares the time window waveform difference of each time window data segment in the initial high-frequency power waveform data stream, retains the time window data segments with a difference greater than 5, and discards the remaining data segments, thereby obtaining a high-frequency power waveform data stream for abnormal identification.

[0103] Another method is based on sorting and proportion selection. The computer system sorts the time window waveform differences of all time window data segments in descending order, and then selects the time window data segments with larger differences according to a certain ratio. For example, the computer system sorts the time window waveform differences of all time window data segments in the initial high-frequency power waveform data stream, and then selects the top 30% of the time window data segments with the largest differences as the high-frequency power waveform data stream to be used for abnormal identification. This method can flexibly adjust the selection ratio according to actual conditions to balance the amount of data and the content of abnormal information.

[0104] When calculating the time window waveform difference, the computer system also needs to consider the division method of the time window. Different time window durations will affect the calculation results of the time window waveform difference. If the time window duration is too short, the data may be too fragmented and cannot accurately reflect the changing trend of the power waveform; if the time window duration is too long, some subtle waveform change information may be obscured. Therefore, the computer system needs to select an appropriate time window duration according to the actual situation. For example, for a faster-changing power waveform, a shorter time window duration, such as 100ms, can be selected; for a relatively slow-changing power waveform, a longer time window duration, such as 1s, can be selected. When selecting a high-frequency power waveform data stream for abnormal identification, historical data and empirical knowledge can also be combined for optimization. By analyzing the data of similar power system failures or normal operation in the past, the relationship between the time window waveform difference and the abnormal situation is summarized. In practical applications, refer to these historical data and empirical knowledge to adjust the time window waveform difference threshold or selection ratio to improve the accuracy and reliability of abnormal identification.

[0105] The computer system can also visualize the selected high-frequency power waveform data stream for abnormal identification. By drawing waveform graphs, difference curves, etc., the changes in the power waveform and the distribution of the waveform difference in the time window can be intuitively displayed. In this way, operators can observe and analyze data more conveniently and discover potential abnormalities in a timely manner.

[0106] As an implementation mode, step S130, according to the time window waveform difference corresponding to each time window data segment in the initial high-frequency power waveform data stream, selects part of the time window data segments in each time window data segment of the initial high-frequency power waveform data stream to obtain the high-frequency power waveform data stream to be used for abnormal identification, including:

[0107] Step S131: obtaining a transient characteristic sequence composed of time window waveform differences corresponding to each time window data segment in the initial high-frequency power waveform data stream;

[0108] Step S132: based on the movement of the dynamic analysis time slot on the transient feature sequence, the time window waveform differences contained in the adjacent movement positions of the dynamic analysis time slot on the transient feature sequence are combined to obtain a plurality of transient combination feature sequences;

[0109] Step S133: For each transient combination feature sequence, based on the feature capture period of the dynamic analysis time slot and the transient combination feature sequence, select some time window data segments from the time window data segments corresponding to the time window waveform difference in the transient combination feature sequence to obtain a high-frequency power waveform data stream for abnormality identification.

[0110] When executing step S131, it is necessary to obtain a transient feature sequence. The time window waveform difference reflects the degree of change of the power waveform between adjacent time window data segments. The time window waveform difference corresponding to each time window data segment in the initial high-frequency power waveform data stream is arranged in chronological order to form a transient feature sequence. The computer system can construct the sequence through the time window waveform difference of each time window data segment calculated in step S120. For example, in the power line monitoring of an electric vehicle charging pile, the initial high-frequency power waveform data stream is divided into 100 time window data segments, and the computer system calculates the time window waveform difference between each time window data segment and the previous adjacent time window data segment as , then the transient characteristic sequence can be expressed as This transient characteristic sequence stores the difference information of electrical quantities in adjacent time windows, providing a basis for subsequent analysis.

[0111] The dynamic analysis time slot is a data segment sliding on a continuous time axis, similar to a time window, for real-time analysis. The computer system moves on the transient feature sequence based on the dynamic analysis time slot, and combines the time window waveform differences contained in the adjacent moving positions covered by the dynamic analysis time slot to obtain multiple transient combination feature sequences. The length of the dynamic analysis time slot determines the number of time window waveform differences contained in each transient combination feature sequence. For example, if the length of the dynamic analysis time slot is set to 5, then when the dynamic analysis time slot is in the transient feature sequence When moving from the first position, the first transient combination characteristic sequence is , when the dynamic analysis time slot moves to the second position, the second transient combination characteristic sequence is , and so on, until the dynamic analysis time slot moves to the end of the transient feature sequence, thus obtaining multiple transient combination feature sequences. In this way, the computer system can analyze the transient feature sequence from different time scales and angles, and capture more detailed information about the changes in the power waveform.

[0112] In step S133, based on the feature capture period of the dynamic analysis time slot and the transient combination feature sequence, some time window data segments are selected from the time window data segments corresponding to the time window waveform difference in the transient combination feature sequence to obtain a high-frequency power waveform data stream to be used for abnormality identification. The feature capture period refers to the time interval for the dynamic analysis time slot to move on the transient feature sequence, which determines the generation frequency of the transient combination feature sequence. The computer system can use a variety of methods to select data segments. One method is based on threshold selection. Specifically, a threshold T is set for the time window waveform difference in the transient combination feature sequence. For each transient combination feature sequence, the time window data segments in which the time window waveform difference is greater than the threshold T are selected. For example, in a transient combination feature sequence In the example, the threshold T is set to 8. If D2=10 and D4=9, the time window data segments corresponding to D2 and D4 are selected.

[0113] Another feasible way is to select based on sorting. Specifically, the time window waveform differences in each transient combination feature sequence are sorted in descending order, and then the time window data segments with larger differences are selected according to a certain ratio. For example, for a transient combination feature sequence, the first 20% of the time window data segments with the largest differences are selected. Assuming that a transient combination feature sequence contains 10 time window waveform differences, the first 2 time window data segments with the largest differences are selected.

[0114] The computer system can also make selections based on the feature capture cycle of the dynamic analysis time slot. If the feature capture cycle is short, it means that the transient combination feature sequence can be obtained more frequently. At this time, the selection ratio can be appropriately increased or the threshold can be lowered to obtain more data segments that may contain abnormal information; if the feature capture cycle is long, in order to avoid selecting too many data segments, the selection ratio can be appropriately reduced or the threshold can be increased. For example, when the feature capture cycle is 100ms, the selection ratio is 30%; when the feature capture cycle is 500ms, the selection ratio is reduced to 20%.

[0115] In order to ensure the accuracy and reliability of the selection, the computer system can also preprocess the transient feature sequence and transient combination feature sequence. Preprocessing includes data smoothing and outlier processing. Data smoothing can use the moving average method. For each time window waveform difference in the transient feature sequence, the average value of several adjacent data points is taken as the smoothed value. For example, for the transient feature sequence , using the 3-point moving average method, the smoothed i-th value (When i is within a reasonable range). Outlier processing can be done by setting upper and lower thresholds, and the time window waveform difference that exceeds the threshold is regarded as an outlier, and then corrected or eliminated. For example, if the upper threshold is set to 20 and the lower threshold is set to 0, if the waveform difference in a time window is 25, it will be corrected to 20; if it is -2, it will be corrected to 0.

[0116] As another implementation, step S100, obtaining a high-frequency power waveform data stream to be used for abnormality identification, includes:

[0117] Step S101: Acquire an initial high-frequency power waveform data stream; the initial high-frequency power waveform data stream is collected when the output power of the charging pile of the target power line undergoes a step change;

[0118] Step S102: determining the signal amplitude distribution statistics corresponding to each time window data segment in the initial high-frequency power waveform data stream;

[0119] Step S103: When it is determined that the initial high-frequency power waveform data stream does not have a noise interference data segment based on the dynamic range compliance analysis results of the high-frequency electrical sampling points in the signal amplitude distribution statistics, the initial high-frequency power waveform data stream is used as the high-frequency power waveform data stream for abnormality identification.

[0120] When executing step S101, the computer system obtains the initial high-frequency power waveform data stream, which records the electrical state changes of the target power line during the step change of the output power of the charging pile. In order to obtain the data stream, the computer system uses various sensors installed on the target power line. The feasible sensors are voltage sensors and current sensors. The voltage sensor can monitor the voltage value in the power line in real time, and the current sensor can monitor the current value. These sensors convert the monitored analog signals into digital signals and transmit them to the computer system. The computer system collects these digital signals at a high-frequency sampling rate to form an initial high-frequency power waveform data stream. For example, in a group of electric vehicle charging piles in a large shopping mall, high-precision voltage and current sensors are installed on the power line of each charging pile, and the computer system collects the data transmitted by these sensors at a sampling frequency of 5000 times per second. When the output power of one of the charging piles changes in a step, the computer system collects the initial high-frequency power waveform data stream reflecting this change process.

[0121] In step S102, the computer system determines the signal amplitude distribution statistics corresponding to each time window data segment in the initial high-frequency power waveform data stream. The time window data segment is obtained by dividing the initial high-frequency power waveform data stream according to a certain time interval, and each time window data segment contains the power waveform data in the time period. The signal amplitude distribution statistics is a statistical analysis of the distribution of the signal amplitude in each time window data segment, which can reflect the frequency of occurrence of the power waveform in different amplitude intervals. Specifically, the histogram statistics method can be used to achieve this purpose. Specifically, the value range of the signal amplitude is first determined, and then this range is divided into several intervals, and the number of signal amplitudes falling in each interval in each time window data segment is counted. For example, for a voltage signal, assuming that its value range is 0V-300V, this range is divided into 10 intervals, namely [0V-30V), [30V-60V), ..., [270V-300V]. For a specific time window data segment, the computer system counts the number of points where the voltage amplitude falls in each interval, thereby obtaining a voltage amplitude distribution histogram corresponding to the time window data segment. By performing such statistics on each time window data segment, the computer system obtains the signal amplitude distribution statistics corresponding to each time window data segment.

[0122] In step S103, the computer system determines whether there is a noise interference data segment in the initial high-frequency power waveform data stream based on the dynamic range compliance analysis results of the high-frequency electrical sampling points in the signal amplitude distribution statistics. The dynamic range of the high-frequency electrical sampling points refers to the range of changes in the signal amplitude within a certain period of time, and the compliance analysis is to determine whether this dynamic range meets the requirements for the normal operation of the power system. If the dynamic range exceeds the normal range, it may mean that there is noise interference. The computer system can set the upper and lower limit thresholds of the dynamic range and compare the amplitude of the high-frequency electrical sampling points in each time window data segment with these thresholds. For example, for the voltage signal, the lower limit threshold of the normal dynamic range is set to 180V and the upper limit threshold is set to 260V. In a certain time window data segment, if there is a voltage sampling point with an amplitude lower than 180V or higher than 260V, and the frequency of occurrence of such abnormal amplitude exceeds a certain proportion, it is considered that there may be noise interference in the time window data segment.

[0123] The computer system can use the following formula to determine whether there is noise interference. Assume that the number of high-frequency electrical sampling points in a certain time window data segment is N, the number of sampling points whose amplitude exceeds the normal dynamic range is n, and the abnormal ratio threshold is set to p. For example, if p=0.1, in a time window data segment containing 100 voltage sampling points, the voltage amplitude of 15 sampling points exceeds the range of 180-260V. , so there is noise interference in the data segment of this time window.

[0124] If the computer system determines through analysis that the initial high-frequency power waveform data stream does not contain a noise interference data segment, the initial high-frequency power waveform data stream is used as the high-frequency power waveform data stream to be subjected to abnormality identification.

[0125] In order to ensure the quality of the initial high-frequency power waveform data stream, the computer system needs to calibrate the sensor regularly when acquiring data. The sensor may have measurement errors during long-term use. By comparing and adjusting with the standard electrical parameter source, the accuracy of the sensor measurement can be guaranteed. For example, the voltage sensor is regularly compared with the standard voltage source to correct the measured value.

[0126] When performing signal amplitude distribution statistics, kernel density estimation can also be used. Kernel density estimation can estimate the probability density function of the signal amplitude more smoothly, thereby more accurately reflecting the distribution of the signal amplitude. Specifically, each sampling point is weighted by a kernel function, and then the weighted results of all sampling points are added to obtain the probability density estimate.

[0127] As an implementation mode, step S300, for each time window data segment in the high-frequency power waveform data stream, determines the waveform feature commonality measurement between the time window data segment and the reference time window data segment, and obtains the waveform commonality measurement index corresponding to the time window data segment, including:

[0128] Step S301: for each time window data segment in the high-frequency power waveform data stream, determine the correlation between the time window data segment and the reference time window data segment, and determine the per-unit value between the time window data segment and the reference time window data segment; the correlation is used to indicate the correlation between the time window data segment and the reference time window data segment; the per-unit value is used to transform the correlation into a preset numerical range;

[0129] Step S302: Based on the ratio of the correlation degree to the per-unit value, determine the waveform feature commonality measurement of the time window data segment and the reference time window data segment, and obtain the waveform commonality measurement index corresponding to the time window data segment.

[0130] When executing step S301, the computer system determines the correlation between the time window data segment and the reference time window data segment. The correlation is an indicator of the correlation between two data segments, which can reflect the similarity of the characteristics of the power waveforms of the two time window data segments. The computer system can use a variety of methods to calculate the correlation, one of which is the correlation coefficient method. The correlation coefficient method determines the correlation by calculating the correlation of the corresponding sampling point values ​​in the two data segments. For example, in the power line monitoring of an electric vehicle charging pile, the reference time window data segment records the voltage waveform when the charging pile is operating normally, and a certain time window data segment records the voltage waveform at a certain moment. The computer system collects the voltage sampling point values ​​of the two data segments and calculates the correlation coefficient r = 0.8 between them using the above formula, which indicates that the voltage waveform of the time window data segment has a strong correlation with the voltage waveform of the reference time window data segment.

[0131] In addition to the correlation coefficient method, the computer system can also use the mutual information method to calculate the correlation, which can measure the dependency between two random variables. In the power waveform analysis, the reference time window data segment and the time window data segment are regarded as two random variables, and the correlation is determined by calculating their mutual information. The larger the mutual information, the stronger the correlation between the two data segments. After determining the correlation, the computer system determines the per-unit value between the time window data segment and the reference time window data segment. The per-unit value is a dimensionless value used to transform the correlation into a preset numerical range for subsequent analysis and comparison. The feasible preset numerical range is [0,1]. The computer system can use a linear transformation method to calculate the per-unit value. Assume that the range of the correlation is , the currently calculated correlation is r, then the per unit value is r pu It can be calculated using the following formula:

[0132] ;

[0133] For example, when the correlation coefficient method is used to calculate the correlation, the range of the correlation coefficient is [-1,1]. If the calculated correlation r = 0.8, then the per-unit value In this way, the correlation is mapped to the interval [0,1], which facilitates the computer system to perform unified processing and comparison.

[0134] In step S302, the computer system determines the waveform feature commonality measurement of the time window data segment and the reference time window data segment based on the ratio of the correlation degree to the per-unit value, and obtains the waveform commonality measurement index corresponding to the time window data segment. The waveform feature commonality measurement is a comprehensive quantitative representation of the degree of similarity of the waveform features of the two time window data segments. The computer system can use the following formula to calculate the waveform commonality measurement index M: ; where r is the correlation, r pu is the per-unit value. The waveform commonality measurement index M can more accurately reflect the commonality of waveform characteristics between the time window data segment and the reference time window data segment. The larger the M value, the higher the commonality of waveform characteristics of the two data segments; the smaller the M value, the lower the commonality of waveform characteristics of the two data segments.

[0135] See also Figure 3 , is a schematic diagram of the structure of a computer system 1000 provided by an embodiment of the present invention. Figure 3 As shown, the above-mentioned computer system 1000 may include: a processor 1001, a network interface 1004 and a memory 1005. In addition, the above-mentioned computer system 1000 may also include: a user interface 1003, and at least one communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. Among them, the user interface 1003 may include a display screen (Display), a keyboard (Keyboard), and the user interface 1003 may optionally include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or it may be a non-volatile memory (non-volatile memory), such as at least one disk storage. The memory 1005 may optionally also be at least one storage device away from the aforementioned processor 1001. As Figure 3 As shown, the memory 1005 as a computer-readable storage medium may include an operating system, a network communication module, a user interface module, and a device control application program.

[0136] exist Figure 3 In the computer system 1000 shown, the network interface 1004 can provide network communication functions; the user interface 1003 is mainly used to provide an input interface; and the processor 1001 can be used to call the device control application stored in the memory 1005 to implement the method provided in the above embodiment.

[0137] It should be understood that the computer system 1000 described in the embodiment of the present invention can execute the above Figure 2 The description of the supercharging host power line protection method in the corresponding embodiment will not be repeated here. In addition, the description of the beneficial effects of using the same method will not be repeated here.

[0138] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

Claims

1. A supercharger host power line protection method, characterized in that: The method comprises: Acquire a high-frequency power waveform data stream for abnormality identification; the high-frequency power waveform data stream is collected when the output power of the charging pile of the target power line changes in a step; Determining a reference time window data segment in each time window data segment of the high-frequency power waveform data stream; For each time window data segment in the high-frequency power waveform data stream, determining a waveform feature commonality metric between the time window data segment and the reference time window data segment, and obtaining a waveform commonality metric index corresponding to the time window data segment; Determine the degree of waveform distortion mutation between adjacent time windows in the high-frequency power waveform data stream according to the waveform commonality measurement index corresponding to each of the time window data segments in the high-frequency power waveform data stream; Determine a fault time window data segment in each of the time window data segments of the high-frequency power waveform data stream based on the waveform distortion mutation degree between the adjacent time windows; the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment of the fault time window data segment is the largest; When it is detected that the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment of the fault time window data segment is greater than the preset protection threshold, the solid-state circuit breaker of the supercharging host is triggered to perform a tripping action.

2. The method according to claim 1, characterized in that Determining the degree of waveform distortion mutation between adjacent time windows in the high-frequency power waveform data stream according to the waveform commonality measurement index corresponding to each time window data segment in the high-frequency power waveform data stream comprises: Modeling transient waveform characteristics according to waveform commonality metric indicators corresponding to each of the time window data segments in the high-frequency power waveform data stream, and obtaining waveform characteristic commonality metric evolution trajectory corresponding to the high-frequency power waveform data stream; The method of determining a fault time window data segment in each of the time window data segments of the high-frequency power waveform data stream based on the degree of waveform distortion mutation between the adjacent time windows comprises: Perform differential feature extraction on the waveform feature commonality metric evolution trajectory to obtain a waveform commonality transient change rate curve corresponding to the high-frequency power waveform data stream; The time window data segment corresponding to the mutation extreme point of the waveform common transient change rate curve is used as the fault time window data segment determined in each of the time window data segments of the high-frequency power waveform data stream; wherein the mutation extreme point is used to indicate that the waveform distortion mutation value between the fault time window data segment and the previous adjacent time window data segment of the fault time window data segment is the largest.

3. The method according to claim 1, characterized in that The fault time window data segment includes a fault triggering time window data segment; the fault triggering time window data segment is a time window data segment triggered by a fault; The step of determining a reference time window data segment in each time window data segment of the high-frequency power waveform data stream comprises: Determine a pre-fault reference time window data segment in each of the time window data segments of the high-frequency power waveform data stream, wherein the pre-fault reference time window data segment is a time window data segment in the high-frequency power waveform data stream that is before the step change and is at the trigger edge at the same time; The key monitoring data segment about the fault-sensitive area in the pre-failure reference time window data segment is used as the reference time window data segment.

4. The method according to claim 1, characterized in that: The fault time window data segment includes a start time window data segment; the step change process includes a step start moment; the start time window data segment is a time window data segment reaching the step start moment; The step of determining a reference time window data segment in each time window data segment of the high-frequency power waveform data stream comprises: Determine a pre-fault reference time window data segment in each of the time window data segments of the high-frequency power waveform data stream, and use the pre-fault reference time window data segment as a reference time window data segment; The pre-fault reference time window data segment is a time window data segment in the high-frequency power waveform data stream that is before the step change and is at the edge of the step.

5. The method according to claim 1, characterized in that The fault time window data segment includes a termination time window data segment; the step change process includes a step termination moment; the termination time window data segment is a time window data segment reaching the step termination moment; The step of determining a reference time window data segment in each time window data segment of the high-frequency power waveform data stream comprises: Determine a step termination time window data segment in each of the time window data segments of the high-frequency power waveform data stream, and use the step termination time window data segment as a reference time window data segment; wherein the step termination time window data segment is a time window data segment in the high-frequency power waveform data stream after the step change is terminated.

6. The method according to claim 5, characterized in that The high-frequency power waveform data stream is obtained by stopping collection after the step change ends; the step end time window data segment includes the last time window data segment in the high-frequency power waveform data stream; The step of determining a step termination time window data segment in each of the time window data segments of the high-frequency power waveform data stream, and using the step termination time window data segment as a reference time window data segment, comprises: The last time window data segment is determined in each of the time window data segments of the high-frequency power waveform data stream, and the last time window data segment is used as a reference time window data segment.

7. The method according to claim 1, characterized in that The step of obtaining a high-frequency power waveform data stream for abnormality identification includes: Acquire an initial high-frequency power waveform data stream; the initial high-frequency power waveform data stream is collected when the output power of the charging pile of the target power line undergoes a step change; For each time window data segment in the initial high-frequency power waveform data stream, based on the instantaneous gradient of the electrical quantity between the time window data segment and the previous adjacent time window data segment of the time window data segment, a time window waveform difference corresponding to the time window data segment is obtained; According to the time window waveform differences corresponding to each of the time window data segments in the initial high-frequency power waveform data stream, some time window data segments are selected from each of the time window data segments in the initial high-frequency power waveform data stream to obtain a high-frequency power waveform data stream for abnormality identification; wherein the time window waveform differences corresponding to the selected time window data segments are greater than the time window waveform differences corresponding to the remaining time window data segments; the remaining time window data segments are the time window data segments in the initial high-frequency power waveform data stream except the selected time window data segments.

8. The method according to claim 7, characterized in that According to the time window waveform difference corresponding to each of the time window data segments in the initial high-frequency power waveform data stream, some time window data segments are selected from each of the time window data segments in the initial high-frequency power waveform data stream to obtain a high-frequency power waveform data stream for abnormality identification, including: Acquire a transient characteristic sequence composed of time window waveform differences corresponding to each of the time window data segments in the initial high-frequency power waveform data stream; Based on the movement of the dynamic analysis time slot on the transient feature sequence, the time window waveform differences contained in adjacent movement positions of the dynamic analysis time slot on the transient feature sequence are combined to obtain a plurality of transient combination feature sequences; For each transient combination feature sequence, based on the feature capture period of the dynamic analysis time slot and the transient combination feature sequence, some time window data segments are selected from the time window data segments corresponding to the time window waveform differences in the transient combination feature sequence to obtain a high-frequency power waveform data stream for abnormality identification.

9. The method according to claim 1, characterized in that: The step of obtaining a high-frequency power waveform data stream for abnormality identification includes: Acquire an initial high-frequency power waveform data stream; the initial high-frequency power waveform data stream is collected when the output power of the charging pile of the target power line undergoes a step change; Determine the signal amplitude distribution statistics corresponding to each time window data segment in the initial high-frequency power waveform data stream; When it is determined that the initial high-frequency power waveform data stream does not have a noise interference data segment according to the dynamic range compliance analysis result of the high-frequency electrical sampling point in the signal amplitude distribution statistics, the initial high-frequency power waveform data stream is used as the high-frequency power waveform data stream to be subjected to abnormality identification; For each time window data segment in the high-frequency power waveform data stream, determining the waveform feature commonality measurement between the time window data segment and the reference time window data segment, and obtaining the waveform commonality measurement index corresponding to the time window data segment, includes: For each time window data segment in the high-frequency power waveform data stream, determine the correlation between the time window data segment and the reference time window data segment, and determine the per-unit value between the time window data segment and the reference time window data segment; the correlation is used to indicate the correlation between the time window data segment and the reference time window data segment; the per-unit value is used to transform the correlation into a preset numerical range; Based on the ratio of the correlation degree to the per-unit value, a waveform feature commonality measurement between the time window data segment and the reference time window data segment is determined to obtain a waveform commonality measurement index corresponding to the time window data segment.

10. A computer system, characterized in that: include: processor; and a memory, wherein the memory stores a computer-readable code, and when the computer-readable code is executed by the processor, the processor executes the method as claimed in any one of claims 1 to 9.

Images