Power flow identification method, system and device based on voltage clustering and storage medium

Abstract

The invention discloses a power flow identification method, system and device based on voltage clustering and a storage medium, and relates to the technical field of electric vehicle charging station topology power flow identification, and the method comprises the steps: collecting voltage data of a charging station distribution network, and sequentially carrying out data integration, abnormal data detection, missing data detection and data consistency check, generating a feature matrix; inputting the generated feature matrix into a hard C-mean clustering algorithm and a fuzzy C-mean clustering algorithm at the same time, grouping charging pile nodes, and calculating the consistency of results of the two algorithms; based on the hard C-means clustering result, the fuzzy C-means clustering result and the consistency index, judging the operation state of the system, and inferring and constructing a final charging station topological structure in combination with the radial characteristics of the power distribution network; according to the invention, the operation efficiency of a power grid is improved, reasonable distribution and intelligent scheduling of charging resources are realized, and intelligent development of a charging network is promoted.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle charging station topology power flow identification technology, and in particular to a power flow identification method, system, device and storage medium based on voltage clustering. Background Technology

[0002] With the rapid growth of electric vehicle ownership and the continuous expansion of charging infrastructure, the charging station network is becoming increasingly complex. In order to improve the efficiency of power grid operation, optimize the allocation of charging resources, and achieve intelligent scheduling, how to accurately grasp the connection relationship between charging stations and charging piles has become a key technical issue.

[0003] Current mainstream topology identification technologies have significant drawbacks: for example, excessive human intervention leads to inefficiency, the computational load of the algorithms is too large to meet real-time requirements, and their adaptability to new charging infrastructure and dynamic power grid environments is insufficient. Specifically, most mainstream topology identification technologies do not perform data preprocessing, resulting in a significant decrease in identification accuracy; they have poor adaptability to dynamic networks or nonlinear relationships, and cannot accurately identify changes in topology; they also struggle to uniformly process data of different scales or sources (such as mixed vector and raster data), and topological relationships may be lost or contradicted due to scale changes; these limitations severely restrict the intelligent development of charging networks. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention provides a power flow identification method, system, device and storage medium based on voltage clustering.

[0005] Therefore, the technical problem solved by this invention is: how to improve the accuracy and dynamic adaptability of topology power flow identification for electric vehicle charging stations, while effectively handling the consistency of multi-source heterogeneous data, and overcoming the problems of low identification accuracy and low efficiency caused by insufficient data preprocessing, poor dynamic network adaptability, and difficulty in uniformly processing multi-scale data in existing methods.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a power flow identification method based on voltage clustering, comprising: The voltage data of the charging station's power distribution network is collected, and the data is integrated, abnormal data is detected, missing data is detected, and data consistency is checked in sequence to generate a feature matrix. The generated feature matrix is ​​simultaneously input into two clustering algorithms, hard C-means and fuzzy C-means, to group the charging pile nodes and calculate the consistency of the results of the two algorithms. Based on the results of hard C-means clustering, fuzzy C-means clustering, and consistency indices, the system operating status is determined, and combined with the radial characteristics of the distribution network, the final charging station topology is inferred and constructed.

[0007] As a preferred scheme for power flow identification based on voltage clustering, wherein: The process involves collecting voltage data from the charging station's power distribution network, performing data integration, anomaly detection, missing data detection, and data consistency checks in sequence, and generating a feature matrix, including: Data integration unifies data from different sources, including time-series data and geographic location information, into a single framework, ensuring that the data format and time series are consistent. Abnormal data that does not conform to the actual usage mode of charging piles is removed through abnormal data detection; The missing data detection method identifies potential data gaps in the transformer substation's main meter and the charging pile's metering. Data consistency checks ensure logical consistency of data. The voltage data is sampled using a sliding window technique, and statistical features are calculated to generate a feature matrix.

[0008] As a preferred scheme for power flow identification based on voltage clustering, wherein: The process of simultaneously inputting the generated feature matrix into both hard C-means and fuzzy C-means clustering algorithms to group charging pile nodes and calculating the consistency of the results from the two algorithms includes: Based on the features extracted by the sliding window, richer features including time-domain features, frequency-domain features, and spatiotemporal correlation features are further extracted. Among them, time-domain features are used to capture the overall trend and stability of the voltage curve, frequency-domain features are extracted by transforming the voltage signal to extract the main frequency components and amplitudes, and spatiotemporal correlation features describe the lag and correlation of charging piles in time and space. The feature values ​​of each charging pile are arranged into row vectors and combined into a two-dimensional feature matrix, where each row represents a sample and each column corresponds to a feature attribute.

[0009] The beneficial effects of this preferred technical solution are as follows: it further extracts time-domain, frequency-domain, and spatiotemporal correlation features, enriching the information dimensions of the feature matrix. Time-domain features can capture the overall trend and stability of the voltage curve, frequency-domain features can extract the frequency components of the voltage signal, and spatiotemporal correlation features can describe the spatiotemporal relationship of the charging pile, thereby more comprehensively and accurately reflecting the operating status of the charging pile and improving the accuracy and effectiveness of clustering.

[0010] As a preferred scheme for power flow identification based on voltage clustering, wherein: The step of simultaneously inputting the generated feature matrix into both hard C-means and fuzzy C-means clustering algorithms to group charging pile nodes and calculating the consistency of the results from the two algorithms also includes: The constructed feature matrix is ​​simultaneously input into two clustering algorithms, hard C-means and fuzzy C-means, to group the charging pile nodes. In hard C-means clustering, the membership matrix and cluster centers are first initialized. The cluster centers are updated according to the cluster center iteration formula, and the membership matrix is ​​updated according to the membership iteration formula. The objective function is calculated. The iteration is continued until the clustering is completed and the samples completely belong to a certain cluster. In fuzzy C-means clustering, given the number of clusters, weighting exponent, and iteration termination condition, the membership matrix and cluster centers are initialized, updated, and the objective function is calculated. The iteration continues by judging whether the error between two iterations of the objective function is less than the iteration termination condition. Samples belong to different clusters to a certain extent.

[0011] The beneficial effects of this preferred technical solution are as follows: It simultaneously employs both hard C-means and fuzzy C-means clustering algorithms to group charging pile nodes, fully leveraging the advantages of both algorithms. Hard C-means clustering ensures that a sample belongs entirely to a single cluster, facilitating clear classification; fuzzy C-means clustering allows samples to belong to different clusters to a certain extent, better handling the fuzziness and uncertainty of the data. Combining the two algorithms improves the reliability and accuracy of the clustering results.

[0012] As a preferred scheme for power flow identification based on voltage clustering, wherein: The step of simultaneously inputting the generated feature matrix into both hard C-means and fuzzy C-means clustering algorithms to group charging pile nodes and calculating the consistency of the results from the two algorithms also includes: Define a consistency index, which is calculated as the ratio of consistent cluster data to total data multiplied by 100%.

[0013] As a preferred scheme for power flow identification based on voltage clustering, wherein: The process of determining the system's operating status based on hard C-means clustering results, fuzzy C-means clustering results, and consistency indices, and inferring and constructing the final charging station topology by combining the radial characteristics of the distribution network, includes: Based on the results of hard C-means clustering and fuzzy C-means clustering, the system's operating status is determined according to the core state identification rules: if the maximum membership degree of a sample in the fuzzy C-means clustering exceeds a certain level and the intra-cluster distance of the hard C-means clustering is less than the threshold 1, it is determined to be in a normal state; if the maximum membership degree is in a specific interval or there are multiple high-membership clusters, it is determined to be in an early warning state; if the inter-cluster distance of the hard C-means clustering in outlier detection is greater than the threshold 2, it is determined to be in an abnormal state.

[0014] As a preferred scheme for power flow identification based on voltage clustering, wherein: The process of determining the system operating status based on hard C-means clustering results, fuzzy C-means clustering results, and consistency indices, and inferring and constructing the final charging station topology by combining the radial characteristics of the distribution network, also includes: The preliminary judgment results are verified based on the consistency index: if the consistency index is higher than the set upper limit, the system operation status judgment result obtained from the core status identification rule is directly adopted; if the consistency index is in the set middle range, manual review is initiated; if the consistency index is lower than the set lower limit, data quality check is triggered.

[0015] The beneficial effects of this preferred technical solution are as follows: Preliminary judgment results are verified based on consistency indicators, allowing for different processing measures to be taken depending on the consistency level. When consistency is high, the judgment result is directly adopted, improving efficiency; when consistency is in the intermediate range, manual review is initiated to correct potential errors; when consistency is low, data quality checks are triggered to ensure data reliability, thereby further improving the accuracy and reliability of system operation status judgment.

[0016] Secondly, the present invention provides a power flow identification system based on voltage clustering, comprising: The data preprocessing and feature generation module is used to collect voltage data from the charging station distribution network, and sequentially perform data integration, abnormal data detection, missing data detection, and data consistency checks to generate a feature matrix. The dual clustering analysis and consistency calculation module is used to simultaneously input the generated feature matrix into two clustering algorithms, hard C-means and fuzzy C-means, to group the charging pile nodes and calculate the consistency of the results of the two algorithms. The operation status judgment module is used to judge the system operation status based on the hard C-means clustering results, fuzzy C-means clustering results and consistency index, and infer and construct the final charging station topology by combining the radial characteristics of the distribution network.

[0017] Thirdly, the present invention provides a computer device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the voltage clustering-based power flow identification method.

[0018] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of a voltage clustering-based power flow identification method.

[0019] The beneficial effects of this invention are as follows: By collecting voltage data, preprocessing, clustering, and topology identification, this invention improves the accuracy of topology relationship identification, enabling a more precise understanding of the connection relationship between charging stations and charging piles, and providing a reliable basis for power grid operation. The provided dynamic update mechanism can adjust the topology diagram in real time to adapt to the dynamic changes in the charging station network, thereby ensuring the stable operation of the power grid. The use of fuzzy clustering of voltage curves for topology identification solves the problem of difficulty in uniformly processing data from different sources, ensuring stable and reliable identification results. This invention helps improve the efficiency of power grid operation, realize the rational allocation and intelligent scheduling of charging resources, and promote the intelligent development of charging networks. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is an overall flowchart of a power flow identification method based on voltage clustering provided by the present invention.

[0022] Figure 2 This is a topology diagram in a simulation example of a power flow identification method based on voltage clustering provided by this invention. Detailed Implementation

[0023] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0024] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a power flow identification method based on voltage clustering, including: S1: Collect voltage data from the charging station's power distribution network, and sequentially perform data integration, abnormal data detection, missing data detection, and data consistency checks to generate a feature matrix; S2: Input the generated feature matrix into both hard C-means and fuzzy C-means clustering algorithms to group the charging pile nodes and calculate the consistency of the results of the two algorithms. S3: Based on the results of hard C-means clustering, fuzzy C-means clustering, and consistency index, determine the system operating status, and combine the radial characteristics of the distribution network to infer and construct the final charging station topology.

[0025] It should be noted that through steps S1-S3, a complete, self-consistent, and reliable charging station topology identification and status awareness system was constructed. Data governance laid a solid foundation for analysis, and the complementary advantages of hybrid clustering models effectively improved the credibility and robustness of the grouping results. Finally, the data-driven clustering relationships were transformed into a power grid topology structure and operational status assessment with clear physical meaning. This not only achieved high-precision and automated identification of charging network connection relationships, reducing reliance on human experience, but also dynamically detected system anomalies, providing key technical support for intelligent operation and maintenance of charging facilities and collaborative optimization of the power grid.

[0026] Example 2, refer to Figure 1 As one embodiment of the present invention, based on the previous embodiment, a power flow identification method based on voltage clustering is provided, including: In this embodiment, step S1 above involves collecting voltage data from the charging station's distribution network and sequentially performing data integration, abnormal data detection, missing data detection, and data consistency checks to generate a feature matrix, including: The voltage data of the charging station's distribution network includes the voltage data of the charging station's transformer substation and charging piles; The voltage data of the charging station distribution network is obtained from the vehicle network platform and the electricity information collection system, and the data is integrated, abnormal data is detected, missing data is detected, and data consistency is checked in sequence. Specifically, data integration involves collecting data from different sources, including time-series data and geographic location information, and integrating this data into a unified data framework to ensure consistency in data format and time series.

[0027] In another possible implementation, data integration can also use ETL (Extract, Transform, Load) tools to extract voltage data from different data sources (such as sensors, smart meters, etc.), clean and transform the data, unify the data format and timestamp, and then load it into a data warehouse for storage and management.

[0028] In another possible implementation, data integration can also be achieved by writing Python scripts that call relevant database interfaces and data processing libraries to read, process, and integrate data from different sources. The scripts can filter, sort, and merge data based on its characteristics and rules, ensuring data consistency and integrity.

[0029] Anomaly detection: Remove abnormal data from the original data that does not conform to the actual usage mode of the charging pile, thereby improving data quality.

[0030] Missing data detection: Detects potential data loss issues in the main meter of the transformer substation and the metering of charging piles, such as meter data at the start, end, and transition stages of charging.

[0031] Data consistency check: Ensures that the data is logically consistent. For example, whether the voltage data after data processing is from the same time.

[0032] Furthermore, after obtaining the voltage data of the charging station's power distribution network, the following operations are also required: Time series alignment: Due to slight time discrepancies in data acquisition from different charging piles and transformer substations, linear interpolation is first used to align the voltage data along the time axis, ensuring that the voltage data of each node corresponds consistently at the same moment. For example, for data with a sampling interval of 15 minutes, if a charging pile is missing a data point between 9:00 and 9:15, it can be filled in by linear interpolation based on the voltage values ​​of the two preceding and following data points.

[0033] Considering that different charging piles may have different voltage amplitude ranges due to their location and load conditions, the Z-score normalization method is used to map the voltage data to a standard normal distribution range with a mean of 0 and a standard deviation of 1.

[0034] Sliding window feature extraction: To capture the dynamic trends of voltage data, a sliding window technique is introduced. Voltage data is sampled over a time window of a certain length (e.g., 1 hour), and statistical features within each window are calculated, including average voltage, standard deviation of voltage fluctuation, and maximum and minimum voltage difference. These features can more comprehensively reflect the characteristics of the voltage curve and serve as input vectors for subsequent clustering.

[0035] In another possible implementation, when collecting voltage data of the charging station distribution network, high-precision voltage sensors can be installed at each key node of the charging station distribution network to automatically collect voltage data at fixed time intervals (such as every 5 minutes) and send the data to the data processing center in real time through a wireless transmission module.

[0036] In another possible implementation, when collecting voltage data of the charging station's distribution network, smart meters can also be used to collect voltage data. Smart meters have data storage and communication functions and can periodically (e.g., daily) upload the collected voltage data to the server. At the same time, the meters can be set to upload data immediately when abnormal voltage fluctuations are detected.

[0037] In this embodiment, step S2 above involves simultaneously inputting the generated feature matrix into both hard C-means and fuzzy C-means clustering algorithms to group the charging pile nodes and calculating the consistency of the results from the two algorithms, including: Based on the features extracted from the sliding window, further richer features are extracted to construct a feature matrix: Time-domain characteristics: In addition to voltage amplitude, calculate time-domain statistical characteristics such as mean, variance (or standard deviation), maximum and minimum values ​​to capture the overall trend and stability of the voltage curve.

[0038] Frequency domain features: Perform Fourier transform on the voltage signal to extract the main frequency components and their amplitudes as feature vectors.

[0039] Spatiotemporal correlation characteristics: Calculate the voltage difference and ratio of adjacent charging piles at different time points to describe the lag and correlation between charging piles in time and space.

[0040] Matrix representation: The feature values ​​of each charging pile are arranged into row vectors, and the row vectors of all charging piles are combined into a two-dimensional matrix. Each row represents a sample, and each column corresponds to a feature attribute.

[0041] The feature matrix generated after the above series of processing steps is simultaneously input into both hard C-means and fuzzy C-means clustering algorithms to group the charging pile nodes, and the consistency of the results of the two algorithms is calculated. The specific operation is as follows: Suppose a sample set contains n samples It is classified into class C, represented as: In the formula, X i (i=1,2,…,c) is the set of the i-th load classification.

[0042] Hard C-means clustering (HCM): Step A.1: After obtaining the number of clusters c in the sample set, initialize the membership matrix. Selecting initial cluster centers Given the iteration termination parameter ; Step A.2: Update the cluster centers according to the iterative formula for cluster centers, expressed as: in, This represents the i-th cluster center, which is continuously updated during the clustering process and eventually represents the center position of the i-th cluster. Represents the j-th sample For the i-th cluster center The membership degree, in hard C-means clustering, The value can be 0 or 1, meaning the sample belongs entirely to a certain cluster.

[0043] Step A.3: Update the membership matrix according to the membership iteration formula, expressed as: Step A.4: Calculate the objective function , represented as: In the formula, The membership matrix, For the set of cluster centers, This indicates the j-th sample. For the i-th cluster center The membership degree, the entire clustering process is to optimize the objective function. The process of minimizing the value.

[0044] Step A.5: Determine the iteration error of the objective function between the two iterations. Is it less than the given iteration termination parameter? If not, proceed to Step A.2; if yes, the iteration ends, and clustering is completed based on the membership matrix.

[0045] Fuzzy C-means clustering (FCM): Step B.1: Given the number of clusters c, the weighting index w, and the iteration termination condition. Select initial cluster centers and initialize the membership matrix; Step B.2: Update the membership matrix and cluster centers. The iterative update formula for the membership matrix elements is: in, It is a weighting index that controls the degree of ambiguity of the sample's membership in different clusters.

[0046] The iterative formula for cluster centers is: in, The range of values ​​is This indicates that the samples belong to different clusters to a certain extent.

[0047] Step B.3: Calculate the objective function , represented as: Determine whether the error between two iterations of the objective function is less than the iteration termination condition. If the result is less than 0, clustering terminates; otherwise, proceed to Step B.2.

[0048] Define consistency metrics: CI=(N consistent / N total )×100% Among them, CI is a consistency index used to measure the degree of consistency between hard C-means clustering and fuzzy C-means clustering results; N consistent For consistent clustering data (i.e., the number of data points in the FCM with the largest number of members equal to the number of data points in the HCM with which clusters are assigned), N total For total data, the total number of samples, i.e. the number of charging pile nodes.

[0049] In another possible implementation, charging pile nodes can be grouped based on their power levels, grouping charging piles with similar power outputs together. First, the power data of all charging piles is statistically analyzed to determine the power range. Then, each charging pile is assigned to its corresponding group based on its actual power value.

[0050] In another possible implementation, when grouping charging pile nodes, they can also be grouped according to their geographical location, grouping geographically adjacent charging piles together. GPS positioning information can be used to calculate the distance between charging piles, and clustering algorithms (such as the DBSCAN algorithm) can be used to cluster nearby charging piles into different groups.

[0051] In this embodiment, step S3 above, based on the hard C-means clustering results, fuzzy C-means clustering results, and consistency index, determines the system operating status, and combines this with the radial characteristics of the distribution network to infer and construct the final charging station topology, including: System operating status assessment includes: Core state recognition rules: If the maximum membership degree is greater than 0.8 and the HCM cluster intra-cluster distance is less than the threshold 1, then it is considered to be in a normal state. If the maximum membership degree is within the interval [0.5, 0.8], or if there are multiple high-membership clusters, then it is determined to be a warning state; If the HCM cluster spacing is greater than the threshold 2 in outlier detection, it is considered an abnormal state.

[0052] It should be noted that in practical applications, threshold 1 and threshold 2 need to be adjusted and optimized according to the specific characteristics of the charging station distribution network system, data features, and actual operating conditions in order to improve the accuracy of system operating status judgment.

[0053] FCM-HCM Consistency Verification Rules: High-reliability identification: CI>85%, directly using the system operation status judgment result obtained from the core status identification rules; Medium confidence identification: 60% < CI ≤ 85%, initiate manual review; Low confidence identification: CI ≤ 60%, trigger data quality inspection.

[0054] In another possible implementation, when inferring and constructing the final topology of the charging station in combination with the radial characteristics of the distribution network, according to the unidirectional power supply characteristics of the radial distribution network, starting from the substation bus, the connection relationship between each charging pile and the upper-level node can be determined in sequence according to the current flow direction. By analyzing the change trend and correlation relationship of voltage data, the branch line to which the charging pile belongs is judged, and the topology of the charging station is gradually constructed.

[0055] In another possible implementation, when inferring and constructing the final topology of the charging station in combination with the radial characteristics of the distribution network, the geographical information system (GIS) can also be used to integrate the line of the distribution network and the geographical location information of the charging piles in combination with the radial characteristics of the distribution network. By analyzing the line direction and length, as well as the spatial position relationship between the charging piles, the topology of the charging station is inferred and visually displayed on the GIS map.

[0056] Embodiment 3, the above is a schematic solution of a power flow identification method based on voltage clustering in this embodiment. It should be noted that the technical solution of a power flow identification system based on voltage clustering belongs to the same concept as the technical solution of the above power flow identification method based on voltage clustering. For the details not described in detail in the technical solution of the power flow identification system based on voltage clustering in this embodiment, reference can be made to the description of the technical solution of the above power flow identification method based on voltage clustering.

[0057] This embodiment also provides a power flow identification system based on voltage clustering, including: Data preprocessing and feature generation module, used to collect voltage data of the charging station distribution network, and perform data integration, abnormal data detection, missing data detection, and data consistency check in sequence, and generate a feature matrix; Biclustering analysis and consistency calculation module, used to input the generated feature matrix into both the hard C-means and fuzzy C-means clustering algorithms at the same time, group the charging pile nodes, and calculate the consistency of the results of the two algorithms; Operating state judgment module, used to judge the operating state of the system based on the hard C-means clustering result, fuzzy C-means clustering result, and consistency index, and infer and construct the final topology of the charging station in combination with the radial characteristics of the distribution network.

[0058] This embodiment also provides an electronic device, applicable to the situation of a power flow identification method based on voltage clustering, including: The system includes a memory and a processor. The memory stores computer-executable instructions, and the processor executes these instructions to implement a voltage clustering-based power flow identification method as proposed in the above embodiments.

[0059] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a power flow identification method based on voltage clustering as proposed in the above embodiments.

[0060] The storage medium proposed in this embodiment belongs to the same inventive concept as the voltage clustering-based power flow identification method proposed in the above embodiment. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0061] Example 4, refer to Figure 2 Tables 1-3, along with one embodiment of the present invention, provide a power flow identification method based on voltage clustering. To verify the beneficial effects of the present invention, simulation experiments are conducted for scientific demonstration.

[0062] A pilot application was conducted at a charging station in a certain city. Electricity consumption data of the charging station's transformer substation and charging piles were collected during a certain period on May 16th. The collected data was processed to obtain the following results: Table 1 Data Acquisition and Processing of Charging Station Substations

[0063] Table 2 Data collected and processed by charging piles

[0064] The clustering results were analyzed to obtain different correlation level classifications for each charging pile: Table 3. Results of Correlation Level Classification

[0065] As shown in Table 3, charging stations numbered 6# and 9# are not among the selected charging stations. The topology diagram of this charging station is shown below. Figure 2 As shown.

[0066] In summary, the voltage clustering-based power flow identification method provided by this invention can effectively classify charging piles into relevance levels. As shown in Table 3, most charging piles have high relevance levels, indicating a strong correlation in their power consumption characteristics, which may be related to the overall operation mode and load distribution of the charging station. Charging piles numbered 6# and 9# have a relevance level of 0 and are identified as not belonging to the selected charging stations. This result demonstrates the accuracy of the power flow identification method in distinguishing charging piles with different topologies. These data characteristics and analysis results provide valuable references for the optimized layout, power distribution, and fault diagnosis of charging stations, contributing to improved operational stability and economic efficiency.

[0067] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A power flow identification method based on voltage clustering, characterized in that, include: The voltage data of the charging station's power distribution network is collected, and the data is integrated, abnormal data is detected, missing data is detected, and data consistency is checked in sequence to generate a feature matrix. The generated feature matrix is ​​simultaneously input into two clustering algorithms, hard C-means and fuzzy C-means, to group the charging pile nodes and calculate the consistency of the results of the two algorithms. Based on the results of hard C-means clustering, fuzzy C-means clustering, and consistency indices, the system operating status is determined, and combined with the radial characteristics of the distribution network, the final charging station topology is inferred and constructed.

2. The power flow identification method based on voltage clustering as described in claim 1, characterized in that, The process involves collecting voltage data from the charging station's power distribution network, performing data integration, anomaly detection, missing data detection, and data consistency checks in sequence, and generating a feature matrix, including: Data integration unifies data from different sources, including time-series data and geographic location information, into a single framework, ensuring that the data format and time series are consistent. Abnormal data that does not conform to the actual usage mode of charging piles is removed through abnormal data detection; The missing data detection method identifies potential data gaps in the transformer substation's main meter and the charging pile's metering. Data consistency checks ensure logical consistency of data. The voltage data is sampled using a sliding window technique, and statistical features are calculated to generate a feature matrix.

3. The power flow identification method based on voltage clustering as described in claim 2, characterized in that, The process of simultaneously inputting the generated feature matrix into both hard C-means and fuzzy C-means clustering algorithms to group charging pile nodes and calculating the consistency of the results from the two algorithms includes: Based on the features extracted by the sliding window, richer features including time-domain features, frequency-domain features, and spatiotemporal correlation features are further extracted. Among them, time-domain features are used to capture the overall trend and stability of the voltage curve, frequency-domain features are extracted by transforming the voltage signal to extract the main frequency components and amplitudes, and spatiotemporal correlation features describe the lag and correlation of charging piles in time and space. The feature values ​​of each charging pile are arranged into row vectors and combined into a two-dimensional feature matrix, where each row represents a sample and each column corresponds to a feature attribute.

4. The power flow identification method based on voltage clustering as described in claim 3, characterized in that, The step of simultaneously inputting the generated feature matrix into both hard C-means and fuzzy C-means clustering algorithms to group charging pile nodes and calculating the consistency of the results from the two algorithms also includes: The constructed feature matrix is ​​simultaneously input into two clustering algorithms, hard C-means and fuzzy C-means, to group the charging pile nodes. In hard C-means clustering, the membership matrix and cluster centers are first initialized. The cluster centers are updated according to the cluster center iteration formula, and the membership matrix is ​​updated according to the membership iteration formula. The objective function is calculated. The iteration is continued until the clustering is completed and the samples completely belong to a certain cluster. In fuzzy C-means clustering, given the number of clusters, weighting exponent, and iteration termination condition, the membership matrix and cluster centers are initialized, updated, and the objective function is calculated. The iteration continues by judging whether the error between two iterations of the objective function is less than the iteration termination condition. Samples belong to different clusters to a certain extent.

5. The power flow identification method based on voltage clustering as described in claim 4, characterized in that, The step of simultaneously inputting the generated feature matrix into both hard C-means and fuzzy C-means clustering algorithms to group charging pile nodes and calculating the consistency of the results from the two algorithms also includes: Define a consistency index, which is calculated as the ratio of consistent cluster data to total data multiplied by 100%.

6. The power flow identification method based on voltage clustering as described in claim 5, characterized in that, The process of determining the system's operating status based on hard C-means clustering results, fuzzy C-means clustering results, and consistency indices, and inferring and constructing the final charging station topology by combining the radial characteristics of the distribution network, includes: Based on the results of hard C-means clustering and fuzzy C-means clustering, the system's operating status is determined according to the core state identification rules: if the maximum membership degree of a sample in the fuzzy C-means clustering exceeds a certain level and the intra-cluster distance of the hard C-means clustering is less than the threshold 1, it is determined to be in a normal state; if the maximum membership degree is in a specific interval or there are multiple high-membership clusters, it is determined to be in an early warning state; if the inter-cluster distance of the hard C-means clustering in outlier detection is greater than the threshold 2, it is determined to be in an abnormal state.

7. The power flow identification method based on voltage clustering as described in claim 6, characterized in that, The process of determining the system operating status based on hard C-means clustering results, fuzzy C-means clustering results, and consistency indices, and inferring and constructing the final charging station topology by combining the radial characteristics of the distribution network, also includes: The preliminary judgment results are verified based on the consistency index: if the consistency index is higher than the set upper limit, the system operation status judgment result obtained from the core status identification rule is directly adopted; if the consistency index is in the set middle range, manual review is initiated; if the consistency index is lower than the set lower limit, data quality check is triggered.

8. A power flow identification system based on voltage clustering, using the method described in any one of claims 1 to 7, characterized in that, include: The data preprocessing and feature generation module is used to collect voltage data from the charging station distribution network, and sequentially perform data integration, abnormal data detection, missing data detection, and data consistency checks to generate a feature matrix. The dual clustering analysis and consistency calculation module is used to simultaneously input the generated feature matrix into two clustering algorithms, hard C-means and fuzzy C-means, to group the charging pile nodes and calculate the consistency of the results of the two algorithms. The operation status judgment module is used to judge the system operation status based on the hard C-means clustering results, fuzzy C-means clustering results and consistency index, and infer and construct the final charging station topology by combining the radial characteristics of the distribution network.

9. An electronic device, characterized in that, include: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores computer-executable instructions that, when executed by a processor, implement the steps of the method according to any one of claims 1 to 7.

Images