A method for identifying low voltage distribution network topology

Abstract

This invention discloses a method for topology identification of low-voltage distribution networks, comprising the following steps: S1, preprocessing the collected voltages at all sampling points to achieve data time alignment and differential voltage processing; S2, using the Pearson cross-correlation algorithm to perform pairwise cross-correlation processing on the differential voltages of all sampling points, forming a cross-correlation matrix; S3, based on a dual-threshold decision mechanism, classifying and aggregating the voltage correlation coefficients to form clusters with high voltage similarity; S4, performing set processing on the clusters to form a set of strongly correlated topological branches, thereby achieving topology identification of the low-voltage distribution network. This invention employs a dual-threshold correlation decision mechanism for differential voltages, which can effectively solve the problem of voltage mutations caused by sudden local interference at branch nodes in the actual topology environment of low-voltage distribution networks, and the phenomenon of large fluctuations in the pairwise cross-correlation coefficients of the same branch triggering missed judgments, thus improving the anti-interference capability of topology identification.

Description

Technical Field

[0001] The present invention relates to the technical field of low-voltage distribution network, and in particular to a method for identifying the topology of a low-voltage distribution network. Background Art

[0002] With the massive access of distributed power sources and charging piles, low-voltage distribution networks face challenges such as intermittent voltage over-limit and three-phase imbalance. Accurate topology information is the basis for grid operation and maintenance, and is essential to ensure safe and stable grid operation. In addition, as a link between the grid and users, the safe, reliable and energy-saving operation of the low-voltage distribution network has a profound impact on the safety, reliability and service quality of users' electricity use.

[0003] At present, the topology identification schemes based on electrical characteristics are divided into two categories: (1) Methods based on characteristic current: using the essential characteristics of current flowing along low-impedance power supply lines, accurate identification of substation topology relationships is achieved through concentrators, distribution transformer terminals and other equipment. (2) Methods based on voltage curve characteristics: extracting the most recognizable local characteristic sub-segments in the voltage curve, forming a similarity metric for clustering, and improving the accuracy of topology identification. Methods based on voltage curve characteristics face challenges in voltage data quality, feature extraction and selection, algorithm complexity and efficiency. Poor data quality (such as measurement errors, missing data or outliers) will reduce the accuracy of topology identification. Summary of the invention

[0004] The purpose of the present invention is to solve the technical problem that the accuracy of low-voltage distribution network topology identification in the prior art is not high, and to propose a low-voltage distribution network topology identification method.

[0005] The technical problem of the present invention is solved by the following technical solutions:

[0006] A method for identifying a low voltage distribution network topology comprises the following steps:

[0007] S1. Preprocess the collected voltages of all collection points to achieve data time alignment and voltage differential processing;

[0008] S2, using the Pearson cross-correlation algorithm to perform pairwise cross-correlation processing on the differential voltages of all acquisition points to form a cross-correlation matrix;

[0009] S3, based on the double threshold decision mechanism, classify and aggregate the voltage correlation coefficients to form clusters with high voltage similarity;

[0010] S4. The aggregated clusters are processed together to form a topological branch set with strong correlation, thereby realizing the topological identification of the low-voltage distribution network.

[0011] In some embodiments, the following technical features are also included:

[0012] In step S1, the data of the collected voltage is aligned at the same time, which specifically includes the following steps:

[0013] For all the acquisition points, the time and the two-dimensional alignment of the acquisition points are performed to ensure that each acquisition point has a unified time stamp. For the moments when there is a sudden loss of acquisition voltage, the linear interpolation method is used to fill the missing values ​​to ensure the continuity and integrity of the data.

[0014] In some embodiments, in step S1, the differential processing of the collected voltage specifically includes the following steps:

[0015] The collected voltage after time alignment and exception processing is processed, and the front and back differences of the collected voltages at adjacent times are processed to eliminate the reference voltage and retain the voltage fluctuation trend. This is used as the final topology identification and analysis collected voltage to reduce storage requirements.

[0016] In some embodiments, in step S2, the Pearson cross-correlation algorithm is used to implement the traversal of the pairwise cross-correlation matrix, which specifically includes the following steps:

[0017] The cross-correlation statistics of N acquisition points are traversed, and the upper and lower triangular symmetric structure of the cross-correlation matrix is ​​used to calculate and store the cross-correlation using the upper triangular form. N ,

[0018]

[0019] Among them, ρ i,j is the Pearson cross-correlation coefficient matrix member of the i-th and j-th acquisition points, thereby simplifying the computational complexity and storage requirements.

[0020] In some embodiments, for the differential voltage data X of the acquisition points X and Y, X=(x1, x2, x3·····x T ), Y=(y1,y2,y3······y T ), calculate its average value The Pearson cross-correlation algorithm is used to obtain the Pearson cross-correlation coefficient matrix member ρ X,Y The calculation formula is:

[0021]

[0022] Wherein, T is the total number of voltage collections at the collection point, k is the index of the current collection number, and the value range is 1 to T.

[0023] In some embodiments, in step S3, the dual threshold decision mechanism includes a low threshold CateThr and a high threshold CateThr_Sec, the low threshold CateThr is used as a clustering threshold for similarity decision to achieve clustering processing; the high threshold CateThr_Sec is used as a decision threshold for the acquisition node with ultra-high correlation, and will not participate in the clustering grouping decision subsequently, thereby reducing the complexity of the decision mechanism.

[0024] In some embodiments, step S3 further includes the following steps:

[0025] S31, establishing an initial search set as a set of all collection point elements, and based on the search set, iteratively taking the first collection point in the search set as a benchmark reference and performing aggregation processing;

[0026] S32, assuming that the first acquisition point index is m, adopt the L-type search method, from the cross-correlation matrix VCorr N Read all its correlation coefficients CorrPot:

[0027]

[0028] Among them, c i,j Represents the cross-correlation matrix VCorr N The value of the i-th row and j-th column, the value range of i and j are both 1 to N;

[0029] S33, judging the mutual correlation coefficient CorrPot based on the low threshold CateThr, extracting the mutual correlation coefficient collection point indexes greater than or equal to the threshold, and forming the NGrp-th aggregation cluster GroupCell[NGrp]:

[0030] GroupCell[NGrp]={m,i∈(c m,j ≥CateThr's collection point index)};

[0031] S34, performing difference processing on the collection point index set of the mutual correlation coefficient greater than or equal to the high threshold CateThr_Sec in CorrPot and the search set index, and updating the search set, that is, the collection points with super high similarity will no longer participate in the next round of aggregation grouping determination; the high threshold CateThr_Sec is greater than the low threshold CateThr.

[0032] In some embodiments, in step S4, the collection processing of the aggregation clusters includes the following steps: based on the aggregation cluster GroupCell, searching for aggregation clusters with one or more identical collection points, performing cluster collection processing to form a new set called the branch topology set ToplogyMap, so that there are no identical collection points between the branch topology sets ToplogyMap after the collection processing.

[0033] The beneficial effects of the present invention compared with the prior art include:

[0034] A low-voltage distribution network topology identification method proposed in the present invention can realize a double-threshold correlation decision mechanism based on differential voltage by preprocessing the collected voltages of all collection points, using the Pearson cross-correlation algorithm to traverse the pairwise cross-correlation processing of the differential voltages of all collection points, classifying and aggregating the voltage correlation coefficients based on a double-threshold decision mechanism, and combining the aggregated clusters, etc., and effectively solves the voltage mutation caused by local sudden interference of branch nodes in the actual topology environment of the low-voltage distribution network, and the phenomenon of triggering missed judgment due to large fluctuations in the range of pairwise cross-correlation coefficients of the same branch, thereby improving the anti-interference ability of topology identification, thereby improving the accuracy of low-voltage distribution network topology identification.

[0035] Other beneficial effects of the embodiments of the present invention will be further described below. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A flow chart of low-voltage distribution network topology identification based on aggregate classification of measured differential voltage cross-correlation in an embodiment of the present invention;

[0037] Figure 2 Schematic diagram of voltage fluctuation trend of all acquisition points of topology branch 1 identified in an embodiment of the present invention;

[0038] Figure 3 Schematic diagram of voltage fluctuation trend of all acquisition points of topology branch 2 identified in an embodiment of the present invention;

[0039] Figure 4 Schematic diagram of voltage fluctuation trend of all acquisition points of topology branch 3 identified in an embodiment of the present invention;

[0040] Figure 5 Schematic diagram of voltage fluctuation trend of all acquisition points of topology branch 4 identified in an embodiment of the present invention;

[0041] Figure 6 Schematic diagram of voltage fluctuation trend of all acquisition points of topology branch 5 identified in an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The present invention is further described below with reference to the accompanying drawings and in combination with preferred embodiments. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0043] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0044] Before describing the embodiments of the present invention, the idea of ​​the present invention is introduced as follows:

[0045] Since the data quality of the existing technology for low-voltage distribution network topology identification is not high, it will directly affect the feature extraction of the voltage curve. Inaccurate voltage data may lead to errors in feature parameter extraction, making the similarity measurement inaccurate, thereby affecting the clustering effect and ultimately leading to inaccurate topology identification results. In the topology identification process, it is necessary to ensure the accuracy and integrity of the voltage data to improve the accuracy and reliability of identification. In terms of algorithm complexity and efficiency: Topology identification methods based on voltage curve features usually require complex algorithms to process and analyze large amounts of data. The design and implementation of the algorithm need to take into account the balance between computational efficiency and accuracy, so the quality of voltage data is an important factor in the topology identification of low-voltage distribution networks.

[0046] Example

[0047] refer to Figure 1 The embodiment of the present invention provides a low voltage distribution network topology identification method based on correlation analysis of measured voltage, comprising the following steps:

[0048] S1. Preprocess the collected voltages of all collection points to achieve data time alignment and voltage differential processing. Data time alignment of collected voltage: By performing two-dimensional alignment of time and collection points for all collection points, ensure that each collection point has a unified timestamp. For moments when there is a sudden missing of collected voltage, use linear interpolation to fill in the missing values ​​to ensure data continuity and integrity. The voltage differential processing is: the collected voltage after time alignment and exception processing is subjected to front and back differential processing of the collected voltages at adjacent moments, the reference voltage is eliminated, and the voltage fluctuation trend is retained as the final topology identification and analysis collected voltage to reduce storage requirements.

[0049] S2. Use the Pearson cross-correlation algorithm to perform pairwise cross-correlation processing on the differential voltages of all acquisition points to form a cross-correlation matrix VCorr N The cross-correlation statistics of N acquisition points are traversed, and the upper and lower triangular symmetric structure of the cross-correlation matrix is ​​used to calculate and store the cross-correlation using the upper triangular form. N ,

[0050]

[0051] Among them, ρ i,j is the Pearson cross-correlation coefficient member of the i-th and j-th acquisition points, simplifying the computational complexity and storage requirements.

[0052] S3. Based on the dual-threshold decision mechanism, the voltage correlation coefficients are classified and aggregated to form aggregate clusters with high voltage similarity. In the dual-threshold decision mechanism, the low threshold CateThr is used as the clustering threshold for similarity judgment to achieve clustering processing; the high threshold CateThr_Sec is used as the decision threshold for the acquisition node with ultra-high correlation, and will not participate in the clustering grouping judgment in the future, reducing the complexity of the judgment mechanism. The specific implementation is divided into three steps: S31. Establish an initial search set as a set of all acquisition point elements. Based on the search set, iteratively use the first acquisition point in the search set as a benchmark reference for aggregation processing. S32. Assuming that the index of the first acquisition point is m, first use the L-type search method to obtain the first acquisition point from the cross-correlation matrix VCorr N Read all its mutual correlation coefficients CorrPot; S33, then judge the mutual correlation coefficient CorrPot based on the low threshold CateThr, extract the mutual correlation coefficient collection point index greater than or equal to the threshold, and form the NGrp-th aggregation cluster GroupCell[NGrp]; S34, finally perform difference set processing on the mutual correlation coefficient collection point index set greater than or equal to the high threshold CateThr_Sec in CorrPot and the search set index, and update the search set, that is, the collection points with super high similarity will no longer participate in the next round of aggregation grouping judgment. Here, CateThr_Sec needs to be greater than CateThr.

[0053]

[0054] GroupCell[NGrp]={m,i∈(c m,j ≥CateThr's collection point index)}

[0055] Among them, c i,j Represents the cross-correlation matrix VCorr N The value of the i-th row and j-th column, the value range of i and j are both 1 to N;

[0056] S4. Perform a collection process on the aggregated clusters to form a strongly correlated topological branch set to achieve topological identification of the low-voltage distribution network. Based on the aggregated cluster GroupCell, find aggregated clusters with one or more identical collection points, perform a collection process on the clusters, and form a new set called a branch topology set ToplogyMap. It is required that there are no identical collection points between the branch topology sets ToplogyMap after the collection process.

[0057] Test example

[0058] In this test example, there are Npot (value 107) collection points in the low-voltage distribution network. For the voltage collection of the collection points, a 15-minute collection cycle is adopted, and Nt (value 96) times are collected daily. The collected data is provided by the substation gateway in accordance with the agreed message format. The collection message must contain: measurement point address-measurement point number-data time-data type information. The double thresholds of this test example are set as low threshold CateThr = 0.7 and high threshold CateThr_Sec = 0.95. The specific implementation steps for the topological branch identification of the substation are as follows:

[0059] Step S1: pre-process the collected voltages of all collection points to achieve data time alignment and voltage differential processing.

[0060] The data is aligned at the moment: by aligning all the acquisition points in time and two-dimensionally, it is ensured that each acquisition point has a unified timestamp. For the moment when there is a sudden missing of the collected voltage, the linear interpolation method is used to fill the missing value to ensure the continuity and integrity of the data, and the collected voltage data Vsta stored in that column is obtained, the size is [96*107]. In the actual low-voltage distribution network, the standard voltage is 220V, and its voltage fluctuation needs to meet a certain fluctuation range. In this test example, the voltage fluctuation is ±20V, so the maximum voltage is 240V and the minimum voltage is 200V.

[0061] The voltage difference processing is: performing the front and back difference processing of the adjacent time acquisition voltages by column, and obtaining the differential voltage of each acquisition point Valys[n,m]=Vsta[n+1,m]-Vsta[n,m], where m∈{1,…107} is the acquisition point index, and n∈{1,…,96} is the acquisition time index. In this implementation case, the maximum statistical differential voltage is 14V and the minimum is -13V. The voltage quantization value range is greatly reduced relative to the voltage cross-correlation processing, which reduces the storage requirements and the complexity of the subsequent Pearson correlation calculation.

[0062] Step S2: Use the Pearson cross-correlation algorithm to perform pairwise cross-correlation processing on the differential voltages of all acquisition points to form a cross-correlation matrix VCorr with upper triangular characteristics. N , the specific steps are:

[0063] Step S201, setting the first acquisition point index i=1;

[0064] Step S202, setting the second acquisition point index j=i+1;

[0065] Step S203: extract the differential voltage X of the ith acquisition point as the differential voltage of the ith column in Valys, and extract the differential voltage Y of the jth acquisition point as the differential voltage of the jth column in Valys.

[0066] Step S204: Correlation coefficient based on Pearson cross-correlation algorithm:

[0067] For the differential voltage X of the ith acquisition point and the differential voltage Y of the jth acquisition point, the differential voltage data X=(x1, x2, x3······x T ), Y=(y1,y2,y3······y T ), calculate its average value The Pearson cross-correlation algorithm is used to obtain the Pearson cross-correlation coefficient matrix member ρ X,Y The calculation formula is:

[0068]

[0069] in is the average value of X and Y, T is the total number of voltage acquisitions at the acquisition point, k is the index of the current acquisition number, and its value range is 1 to T. , update the cross-correlation matrix VCorr[i,j]=ρ X,Y Step S205: perform j++ calculation. If j is less than or equal to N, return to step S204. If j is greater than N, proceed to step S206.

[0070] Step S206, perform i++ calculation, and if it is determined that i is less than or equal to N, return to step S202, otherwise, end the update of the mutual correlation matrix;

[0071] Step S3: Based on the dual threshold decision mechanism, the voltage correlation coefficients are classified and aggregated to form an aggregate cluster with high voltage similarity. The specific implementation steps are as follows:

[0072] Step S301, establish an initial search set HvPot as a set of all collection point elements, HvPot = {1, 2, ..., NPot}, and initialize the aggregation cluster index NGrp = 0;

[0073] Step S302: Make a non-empty determination on HvPot. If it is not empty, extract the first acquisition point of HvPot as the basic reference acquisition point AlysPot, and proceed to step S303; otherwise, terminate the aggregation process.

[0074] Step S303: Use L-type search method to find the correlation matrix VCorr N Read all its correlation coefficients CorrPot.

[0075]

[0076] Among them, c i,j Represents the cross-correlation matrix VCorr N The value of the i-th row and j-th column, the value range of i and j are both 1 to N;

[0077] Step S304: judge the mutual correlation coefficient CorrPot based on the low threshold CateThr, extract the mutual correlation coefficient collection point index greater than or equal to the threshold, if it exists, perform NGrp++, and form the element set of the corresponding collection point index into the aggregation cluster GroupCell[NGrp], and enter step S305 for processing, otherwise return to step S302;

[0078] Step S305: judge the mutual correlation coefficient CorrPot based on the high threshold CateThr_Sec, extract the mutual correlation coefficient collection point index greater than or equal to the threshold as the set SelectPot, if it exists, update the search set HvPot = HvPot-SelectPot, that is, perform difference processing on SelectPot and the search set HvPolt, and return to step S302.

[0079] Step S4: aggregate the aggregated clusters to form a topological branch set with strong correlation, thereby realizing the topological identification of the low-voltage distribution network. The specific implementation steps are as follows:

[0080] Step S401, create and initialize all zeros of the aggregation cluster merge flag set ComFlag[NGrp,NGrp];

[0081] Step S402, traverse and search for clusters, if clusters i and j have one or more common collection points, set ComFlag[i, j] to 1, where i ranges from 1 to NGrp, and j ranges from i to NGrp;

[0082] Step S403: Based on the merging flag set ComFlag, the aggregation cluster GroupCell is processed to complete the branch topology identification of the low-voltage distribution network. The specific implementation steps are as follows:

[0083] Step S4031, initializing cluster index ii=1;

[0084] Step S4032: If ii is not greater than NGrp, initialize the row index RIdx_1 = ii, and initialize the number of collections LL = 0, otherwise proceed to step S5;

[0085] Step S4033, perform row search: find all column indexes of 1 in the ii-th row of ComFlag as a set CIdx, count their number LL, and set the elements of the ii-th row of ComFlag whose value is 1 to 0;

[0086] Step S4034: If LL is 0, return to step S4032, otherwise go to step S4035;

[0087] Step S4035, conduct column search: use all elements in CIdx as column indexes, search for the row index stored as 1 in the corresponding column in ComFlag, record it as CombPot, and set the ComFlag at the corresponding position to 0;

[0088] Step S4036, deduplicate the elements in CombPot to form a new set RIdx_2;

[0089] Step S4037: All elements in RIdx_2 that do not belong to RIdx_1 are taken as all members of RIdx_3, and the statistical length is Nr3;

[0090] Step S4038: If RIdx_3 is not empty, perform the following processing; otherwise, proceed to step S4039.

[0091] (a) Merge all elements in RIdx_3 into RIdx_1 and set RIdx_2 to empty;

[0092] (b) Using all elements in RIdx_3 as row search indexes, find the column index corresponding to 1 in ComFlag, record it as CIdx_2, the statistical length is Nc2, and set the ComFlag at the corresponding position to 0;

[0093] (c) Using all elements in CIdx_2 as column search indexes, find the row index stored as 1 in the corresponding column in ComFlag, write it into the set RIdx_2, and set the ComFlag at the corresponding position to 0;

[0094] (d) De-duplicate the elements in RIdx_2 to form a new set RIdx_2;

[0095] (e) All elements in RIdx_2 that do not belong to RIdx_1 are taken as all members of RIdx_3, and the statistical length is Nr3, and the process returns to the judgment process of step S4038.

[0096] Step S4039: deduplicate the elements in RIdx_1, update it to CombPotSet, and count the number of elements Nset.

[0097] Step S40310: Using the element in CombPotSet as the index of the cluster, merge all the collection points in the corresponding cluster, and perform deduplication processing to form all members of the topology branch set TopologyMap[Nsub], and return to step S4032.

[0098] Step S5, ending the aggregation processing of the aggregated clusters, and outputting the final topology result TopologyMap of the low-voltage distribution network branch.

[0099] In this test example, the low-voltage distribution network topology identification method based on the aggregate classification of the differential voltage cross-correlation measurement according to the embodiment of the present invention is adopted to identify a total of 5 topology branches SubNet1 to SubNet5. Figure 2-6 The collection point index of each branch and the voltage fluctuation trend of the branch collection point are given. It can be seen from the figure that the overall trend of the collection point voltage of each branch is basically the same, with a high correlation, which is consistent with the same phase of the branch in the same substation, and its voltage characteristics and voltage fluctuations are similar. Figure 3 The topology branch SubNet2 found that the collected voltage at the collection point in the branch had local voltage mutations and inconsistent trends, which can be considered as voltage mutations introduced by unstable voltage quality or branch load changes. This test case uses double threshold judgment to solve the branch missed detection caused by voltage mutations. It has good anti-interference ability and improves the accuracy of low-voltage distribution network topology identification.

[0100] Compared with the prior art, the embodiments of the present invention have the following advantages:

[0101] 1. Use differential voltage processing: On the one hand, for time-varying circuit loads, the branch voltage is a fluctuating voltage of 220V±delta, and differential voltage processing is used. The actual storage value range is ±2delta, which reduces the quantization bit width and thus reduces storage resources. On the other hand, Pearson correlation processing is performed on the differential voltage, which reduces the processing data bit width of the correlation processing and reduces the processing resource requirements for calculation. Secondly, differential voltage is used for correlation processing to extract voltage fluctuation characteristics. The distribution of correlation coefficients presents a more obvious high similarity advantage, which reduces the requirements for threshold setting and can better adapt to environmental changes.

[0102] 2. A dual-threshold decision mechanism is adopted: the low threshold is used as the clustering threshold for similarity judgment to realize clustering processing; the high threshold optimizes the search set to reduce the complexity of the decision mechanism; at the same time, the dual-threshold decision can effectively solve the voltage mutation caused by local sudden interference at branch nodes in the actual topology environment of the low-voltage distribution network, the voltage quality degradation, and the large fluctuation of the mutual correlation coefficient between the same branch and the triggering of missed judgment, thereby improving the anti-interference ability of topology identification, thereby improving the accuracy of low-voltage distribution network topology identification.

[0103] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art of the present invention, several equivalent substitutions or obvious variations can be made without departing from the concept of the present invention, and the performance or use is the same, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. A method for identifying low voltage distribution network topology, characterized in that: The steps include: S1. Preprocess the collected voltages of all collection points to achieve data time alignment and voltage differential processing; S2, using the Pearson cross-correlation algorithm to perform pairwise cross-correlation processing on the differential voltages of all acquisition points to form a cross-correlation matrix; S3, based on the double threshold decision mechanism, classify and aggregate the voltage correlation coefficients to form clusters with high voltage similarity; S4. The aggregated clusters are processed together to form a topological branch set with strong correlation, thereby realizing the topological identification of the low-voltage distribution network.

2. A low voltage distribution network topology identification method according to claim 1, characterized in that: In step S1, the data of the collected voltage is aligned at the same time, which specifically includes the following steps: For all the acquisition points, the time and the two-dimensional alignment of the acquisition points are performed to ensure that each acquisition point has a unified time stamp. For the moments when there is a sudden loss of acquisition voltage, the linear interpolation method is used to fill the missing values ​​to ensure the continuity and integrity of the data.

3. A low voltage distribution network topology identification method according to claim 1, characterized in that: In step S1, the differential processing of the collected voltage specifically includes the following steps: The collected voltage after time alignment and exception processing is processed, and the front and back differences of the collected voltages at adjacent times are processed to eliminate the reference voltage and retain the voltage fluctuation trend. This is used as the final topology identification and analysis collected voltage to reduce storage requirements.

4. A low voltage distribution network topology identification method according to claim 1, characterized in that: In step S2, the Pearson cross-correlation algorithm is used to implement the traversal of the pairwise cross-correlation matrix, which specifically includes the following steps: The cross-correlation statistics of N acquisition points are traversed, and the upper and lower triangular symmetric structure of the cross-correlation matrix is ​​used to calculate and store the cross-correlation using the upper triangular form. N , Among them, ρ i,j is the Pearson cross-correlation coefficient matrix member of the i-th and j-th acquisition points, thereby simplifying the computational complexity and storage requirements.

5. A low voltage distribution network topology identification method according to claim 4, characterized in that: For the differential voltage X of the ith acquisition point and the differential voltage Y of the jth acquisition point, the differential voltage data X=(x1, x2, x3······x T ), Y=(y1,y2,y3······y T ), calculate its average value The Pearson cross-correlation algorithm is used to obtain the Pearson cross-correlation coefficient matrix member ρ X,Y The calculation formula is: Wherein, T is the total number of times the voltage of the acquisition point is collected, k is the index of the current acquisition number, and the value range is 1 to T.

6. A low voltage distribution network topology identification method according to claim 1, characterized in that: In step S3, the dual threshold decision mechanism includes a low threshold CateThr and a high threshold CateThr_Sec, and the low threshold CateThr is used as a clustering threshold to perform similarity decision to implement clustering processing; The high threshold CateThr_Sec is used as a decision threshold for the acquisition node to have an ultra-high correlation, and will not participate in clustering grouping decisions later, thereby reducing the complexity of the decision mechanism.

7. A low voltage distribution network topology identification method according to claim 6, characterized in that: Step S3 also includes the following steps: S31, establishing an initial search set as a set of all collection point elements, and based on the search set, iteratively taking the first collection point in the search set as a benchmark reference and performing aggregation processing; S32, assuming that the first acquisition point index is m, adopt the L-type search method, from the cross-correlation matrix VCorr N Read all its correlation coefficients CorrPot: Among them, c i,j Represents the cross-correlation matrix VCorr N The value of the i-th row and j-th column, the value range of i and j are both 1 to N; S33, judging the mutual correlation coefficient CorrPot based on the low threshold CateThr, extracting the mutual correlation coefficient collection point indexes greater than or equal to the threshold, and forming the NGrp-th aggregation cluster GroupCell[NGrp]: GroupCell[NGrp]={m,i∈(c m,j ≥CateThr's collection point index)}; S34, performing difference processing on the collection point index set of the mutual correlation coefficient greater than or equal to the high threshold CateThr_Sec in CorrPot and the search set index, and updating the search set, that is, the collection points with super high similarity will no longer participate in the next round of aggregation grouping determination; the high threshold CateThr_Sec is greater than the low threshold CateThr.

8. A low voltage distribution network topology identification method according to claim 1, characterized in that: In step S4, the aggregation cluster collection processing includes the following steps: based on the aggregation cluster GroupCell, find the aggregation clusters with one or more identical collection points, perform cluster collection processing, and form a new set called the branch topology set ToplogyMap, so that there are no identical collection points between the branch topology sets ToplogyMap after the collection processing.

Images