Low-voltage transformer area topology identification method based on ammeter data

Abstract

The present application relates to the field of data processing, especially to a low-voltage transformer area topology identification method based on electric meter data, comprising: obtaining voltage sequences of each to-be-identified object in a low-voltage transformer area; dividing each voltage sequence according to time periods, and calculating the time and frequency domain discriminant of each to-be-identified object in the same time period; traversing to calculate the time domain distance and frequency domain distance of two to-be-identified objects in each time period, assigning weights to the time domain discriminant and frequency domain discriminant and to the time domain distance and frequency domain distance, and calculating the target distance by weighted fusion, clustering each to-be-identified object based on the target distance, and calculating the membership of each to-be-identified object according to the clustering result; inputting the obtained membership of each to-be-identified object into the system to generate a low-voltage transformer area topology map to complete topology identification. The present topology identification method accurately captures the differences of different to-be-identified objects in the time and frequency domain, and improves the identification accuracy of the membership.

Description

Technical Field

[0001] This invention relates to the field of data processing, and in particular to a method for identifying the topology of low-voltage distribution areas based on electricity meter data. Background Technology

[0002] Low-voltage distribution network topology refers to the connection relationships and layout of devices such as users, meter boxes, and branch boxes in a power system's low-voltage distribution network. Existing technologies typically utilize the similarity of voltage sequences for topology identification. For example, the correlation coefficient between user voltage sequences is calculated to determine whether users belong to the same phase, or clustering algorithms are used to cluster voltage sequences to determine the affiliation between users and meter boxes, and between meter boxes and branch boxes.

[0003] Current technologies, when using the entire voltage sequence for calculations, do not consider the impact of load characteristic variations at different times (such as peak and off-peak electricity consumption periods) on the voltage sequence. During peak electricity consumption periods, the simultaneous start-up and shutdown of various high-power devices generates strong characteristic harmonics, leading to drastic fluctuations in the amplitude of each harmonic and complex changes in time-domain characteristics. In contrast, during off-peak periods, devices are stable and singular, with low harmonic content and obvious changes in time-domain characteristics. Furthermore, current technologies mainly rely on time-domain characteristics (such as voltage amplitude variations), while frequency-domain characteristics have significant distinguishing power when nonlinear loads (such as air conditioners and induction cookers) are started and stopped, resulting in decreased accuracy in identifying complex load periods. Summary of the Invention

[0004] To address the problem that existing topology identification methods, which directly cluster the entire voltage sequence and its feature sequences, ignore the differences in the time and frequency domains during different electricity consumption periods, leading to misidentification of the affiliation of the objects to be identified, this invention provides a voltage distribution area topology identification method based on electricity meter data.

[0005] A low-voltage distribution area topology identification method based on meter data includes: acquiring the voltage sequence of each object to be identified in the low-voltage distribution area; segmenting each voltage sequence according to time periods, calculating the time-domain and frequency-domain distinguishability of each object to be identified in the same time period; traversing and calculating the time-domain and frequency-domain distances of two objects to be identified in each time period, assigning weights to the time-domain and frequency-domain distinguishability as the time-domain and frequency-domain distances, weighted fusing the time-domain and frequency-domain distances to calculate the target distance, clustering each object to be identified based on the target distance, calculating the membership relationship of each object to be identified based on the clustering results; and inputting the obtained membership relationship of each object to be identified into the system to generate a low-voltage distribution area topology map to complete the topology identification.

[0006] By adopting the above technical solutions, the identification accuracy of the affiliation of each device to be identified in the low-voltage distribution area topology is effectively improved. During peak electricity consumption periods, multiple devices start up together, resulting in rich harmonics and complex time-domain fluctuations. During off-peak electricity consumption periods, devices are single and stable with obvious time-domain characteristics. By calculating the distance in segments, the voltage changes during peak and off-peak electricity consumption periods can be dynamically adapted. Purely resistive loads such as electric heaters have large time-domain fluctuations and few harmonics, while nonlinear loads such as variable frequency air conditioners have stable time-domain but rich harmonics. Compared with the traditional Euclidean distance between the entire sequence as a distance metric, when the target distance is calculated based on the time and frequency domain distance and time and frequency domain discrimination of different time periods as a distance metric for clustering, the time and frequency domain two-dimensional distance can accurately capture the dual differences between different objects to be identified, significantly improving the accuracy of identifying the affiliation of different objects to be identified.

[0007] Preferably, the calculation process of the time-domain discriminant includes: Based on the voltage sequences within each time period, the ratio of the Manhattan distance to the maximum Manhattan distance for each object to be identified in the same time period is calculated pairwise, and the sum of the ratios is used as the temporal distinguishability of different objects to be identified in that time period.

[0008] Preferably, the calculation process of the frequency domain discrimination includes: Fourier transform is performed on the voltage sequence of the object to be identified in different time periods to obtain the spectrum sequence in that time period. Based on the spectrum sequence in each time period, the ratio of the Manhattan distance to the maximum Manhattan distance of each object to be identified in the same time period is calculated pairwise, and the sum of the ratios is used as the frequency domain distinguishability of different objects to be identified in that time period.

[0009] By employing the above technical solution, the Manhattan distance ratio in the time domain and frequency domain is calculated respectively. This method can comprehensively capture the differences between the objects to be identified from two dimensions: time series and frequency distribution. The discriminability in the time domain and frequency domain reflects the degree of difference in the voltage series in the time and frequency dimensions, respectively. Differences in the time domain may be reflected in the fluctuation and trend of voltage, while differences in the frequency domain may be reflected in the harmonics and frequency components of voltage. The combination of the two makes the discriminability more comprehensive and accurate.

[0010] Preferably, the calculation process of the time-domain distance includes: Based on the voltage sequences within each time period, the absolute value of the standard deviation difference of the voltage sequences of each object to be identified in the same time period is calculated pairwise to obtain the time-domain fluctuation difference; the absolute value of the difference of each element of the voltage sequence of each object to be identified in the same time period is calculated pairwise and summed to obtain the sum of time-domain element differences; the sum of the time-domain fluctuation difference and the sum of time-domain element differences is used as the time-domain distance between the two objects to be identified in that time period.

[0011] Preferably, the calculation process of the frequency domain distance includes: Perform Fourier transform on the voltage sequences of the objects to be identified within different time periods to obtain the spectral sequences within those time periods. Based on the spectral sequences within each time period, calculate the absolute value of the standard deviation difference of the spectral sequences of each object to be identified in the same time period for each pair of objects to be identified, and obtain the frequency domain fluctuation difference. Calculate the absolute value of the difference of each element of the spectral sequence of each object to be identified in the same time period for each pair of objects to be identified and sum them to obtain the sum of the frequency domain element differences. Sum the frequency domain fluctuation difference and the sum of the frequency domain element differences to obtain the frequency domain distance between the two objects to be identified in that time period.

[0012] By adopting the above technical solutions, time-domain fluctuation differences are more sensitive to purely resistive devices with large fluctuations and low harmonics, while frequency-domain fluctuation differences are more sensitive to nonlinear devices with small fluctuations and high harmonics. Combining the two can accurately distinguish different types of devices and improve the recognition accuracy. The voltage sequences of different time periods are calculated separately in the time and frequency domains, which can dynamically adapt to voltage changes under different operating conditions during peak and off-peak electricity consumption periods, avoid smoothing out time-period differences caused by overall calculation, and ensure recognition accuracy.

[0013] Preferably, the target distance is calculated as follows: The time-domain and frequency-domain discrimination are normalized to obtain the time-domain weight and frequency-domain weight. The product of the time-domain weight and the time-domain distance is used as the time-domain weighted distance, and the product of the frequency-domain weight and the frequency-domain distance is used as the frequency-domain weighted distance. The target distance is obtained by summing the time-domain weighted distance and the frequency-domain weighted distance for each time period.

[0014] Preferably, the target distance is used as a distance metric to cluster all objects to be identified to obtain multiple clusters. The objects to be identified include branch boxes, table boxes, and users. For a cluster, the Kendall coefficient between the user to be identified and each table box to be identified, and between the table box to be identified and each branch box to be identified, is calculated. The object to be identified with the highest correlation of the Kendall coefficient is selected as the member object. One user to be identified belongs to a unique table box to be identified, and one table box to be identified belongs to a unique branch box to be identified.

[0015] By adopting the above technical solution, the normalized time domain and frequency domain distinguishability are weighted, which can dynamically adjust the contribution of the time domain and frequency domain in distance calculation according to data characteristics. Through weighted fusion, the interference of single-dimensional anomalies on clustering results can be effectively reduced. This method can automatically adjust the importance of the time domain and frequency domain according to the characteristics of actual data. Dynamic weights and multi-dimensional fusion can effectively reduce misjudgments caused by single features or unclear membership relationships.

[0016] Preferably, when segmenting voltage sequences according to time periods, the optimal segmentation method is selected, and the specific operation is as follows: The voltage sequence of each object to be identified is divided into multiple voltage subsequences. Fourier transform is used to obtain the spectral sequence of each voltage subsequence in each object to be identified. The standard deviation of the frequency amplitude of each voltage subsequence is calculated based on the spectral sequence. The variance is calculated based on the standard deviation of the frequency amplitude of each voltage subsequence. The optimal segmentation method of the voltage sequence is selected according to the variance.

[0017] The present invention has the following effects: This method effectively improves the accuracy of identifying the affiliation of various devices in the low-voltage distribution area topology. During peak electricity consumption periods, multiple devices start up together, resulting in rich harmonics and complex time-domain fluctuations. During off-peak electricity consumption periods, individual devices exhibit stable time-domain characteristics. By calculating distances in segments, it can dynamically adapt to voltage changes during peak and off-peak electricity consumption periods. Purely resistive loads, such as electric heaters, exhibit large time-domain fluctuations but few harmonics, while nonlinear loads, such as variable frequency air conditioners, exhibit stable time-domain performance but rich harmonics. Compared with the traditional Euclidean distance between the entire sequence as a distance metric, this method calculates the target distance based on the time and frequency domain distances and time and frequency domain discriminative values ​​at different time periods as a distance metric for clustering. The time and frequency domain dual-dimensional distance can accurately capture the dual differences between different objects to be identified, significantly improving the accuracy of identifying the affiliation of different objects. Attached Figure Description

[0018] Figure 1 This is a flowchart of steps S1-S4 in the low-voltage transformer area topology identification method based on meter data according to an embodiment of the present invention.

[0019] Figure 2 This is a schematic diagram showing the segmentation of voltage sequences of different objects to be identified according to time periods in the low-voltage distribution area topology identification method based on meter data in an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0021] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0022] Reference Figure 1 The method for identifying the topology of low-voltage distribution areas based on electricity meter data has the following specific steps: Step S1: Obtain the voltage sequence of each object to be identified in the low-voltage distribution area topology.

[0023] The objects to be identified include users, meter boxes, and branch boxes in the low-voltage topology. The voltage sequences of the users and meter boxes are collected by smart meters, while the voltage sequences of the branch boxes are collected using non-invasive CT sensors. The voltage sequences of each object to be identified are acquired simultaneously at the same time. Wavelet denoising is used to preprocess the acquired voltage sequences, thus obtaining the voltage sequence of each object to be identified.

[0024] Step S2: Divide each voltage sequence into time periods, and calculate the time-domain and frequency-domain discrimination of each object to be identified in the same time period.

[0025] To reflect the differences in the time and frequency domains of the objects to be identified during different power consumption periods, the voltage sequence of the objects to be identified is segmented according to time periods. Taking into account the differences in the time and frequency domains of the objects to be identified during different power consumption periods, the optimal segmentation method of the voltage sequence is selected, and the time domain discrimination and frequency domain discrimination of each object to be identified are calculated pairwise at the same time interval.

[0026] Figure 2 This represents the voltage sequence segmentation method for different objects to be identified, where k1, k2, and k3 represent different time periods, and A, B, and C represent different objects to be identified.

[0027] When selecting the optimal segmentation method for the voltage sequence, the voltage sequence of each object to be identified is first segmented. For ease of description, multiple voltage sequences segmented according to time periods are called voltage subsequences. Specifically, the voltage sequence can be segmented equally in time, or it can be segmented according to the time periods of peak and off-peak electricity consumption.

[0028] Next, the spectrum of the voltage subsequence of each object to be identified is obtained through Fourier transform. The spectrum sequence of each voltage subsequence is then obtained from the spectrum, where each element represents the amplitude of each frequency in the spectrum. The standard deviation of the frequency amplitude of each voltage subsequence is calculated based on the spectrum sequence. The variance of the standard deviations of the frequency amplitudes of all voltage subsequences is calculated as the segmentation score for that object. The average of the segmentation scores for all objects to be identified is then used to obtain the comprehensive score.

[0029] The voltage frequency domain characteristics change differently when users use devices with different power levels. The standard deviation indicates that the difference in amplitude at different frequencies of a voltage subsequence is unique. If there is a difference in the standard deviation of the frequency domain characteristics between two voltage subsequences, it means that there is a difference in the distribution of the frequency domain characteristics between the two voltage subsequences.

[0030] Repeat the above operation to iterate on different segmentation methods, and select the segmentation method with the highest comprehensive score as the optimal segmentation method. The voltage subsequence of the optimal segmentation method reflects the load status of each object to be identified in the low-voltage distribution area.

[0031] The calculation process of time-domain discrimination is as follows: based on the voltage sequence in each time period, calculate the ratio of the Manhattan distance to the maximum Manhattan distance of each object to be identified in the same time period, and sum the ratios as the time-domain discrimination of different objects to be identified in that time period.

[0032] The specific formula for calculating the time-domain discrimination is as follows: ; in, Indicates time period Temporal discriminability, Indicates the number of objects to be identified. Indicates the time period Upper The object to be identified and the first Manhattan distance between voltage subsequences of each object to be identified This represents the maximum Manhattan distance of all voltage subsequences of the objects to be identified.

[0033] The frequency domain discrimination calculation process is as follows: Fourier transform is performed on the voltage sequences of the object to be identified within different time periods to obtain the spectrum diagram for that time period, and the spectrum sequence is obtained from the spectrum diagram. Based on the spectrum sequences within each time period, the ratio of the Manhattan distance to the maximum Manhattan distance for each object to be identified in the same time period is calculated pairwise, and the sum of these ratios is used as the frequency domain discrimination of different objects to be identified in that time period.

[0034] The formula for calculating frequency domain discrimination is as follows: ; in, Indicates time period Frequency domain discrimination Indicates the number of objects to be identified. Indicates the time period Upper The object to be identified and the first Manhattan distance between the spectral sequences of the objects to be identified This represents the maximum Manhattan distance of the spectral sequences of all objects to be identified.

[0035] For different objects to be identified, the time-domain and frequency-domain discriminative values ​​are normalized so that the sum of the time-domain and frequency-domain discriminative values ​​is adjusted to 1. The normalized time-domain discriminative value is used as the time-domain weight, and the normalized frequency-domain discriminative value is used as the frequency-domain weight.

[0036] Because the electricity consumption intensity of users within a distribution area varies at different times, the concentrated start-up and shutdown of multiple high-power devices during peak periods generates strong characteristic harmonics, causing drastic fluctuations in the amplitude of each harmonic and resulting in complex changes in time-domain characteristics. In contrast, during off-peak periods, the devices are generally stable and single, with lower harmonic content, and the changes in time-domain characteristics are more obvious. The degree of differentiation of the low-voltage distribution area topology differs in both the time and frequency domains across different electricity consumption periods.

[0037] By segmenting the voltage sequence and calculating time-domain and frequency-domain discrimination, it is easier to identify the load status of users using equipment at different times. Taking into account the local differences between peak and off-peak electricity consumption periods is more in line with actual usage scenarios. Calculating time-domain and frequency-domain discrimination avoids the limitations of identification using a single time-domain feature. When the time-domain features change complexly during peak periods, combining frequency-domain features for identification can achieve accurate identification of the affiliation of each object.

[0038] Step S3: Iterate through and calculate the temporal and frequency distances of the two objects to be identified in each time period. Assign weights to the temporal and frequency distances based on the temporal and frequency discrimination scores. Calculate the target distance by weighted fusion of the temporal and frequency distances. Cluster each object to be identified based on the target distance. Calculate the membership relationship of each object to be identified based on the clustering results.

[0039] The calculation process of time-domain distance is as follows: Based on the voltage sequence in each time period, calculate the absolute value of the standard deviation difference of the voltage subsequence of each object to be identified in the same time period for each pair of objects to be identified, and obtain the time-domain fluctuation difference; calculate the absolute value of the difference of each element of the voltage subsequence of each object to be identified in the same time period for each pair of objects to be identified and sum them up to obtain the sum of time-domain element differences; sum the time-domain fluctuation difference and the sum of time-domain element differences as the time-domain distance between the two objects to be identified in that time period.

[0040] The specific reference formula for time-domain distance is: ; in, Indicates the first The and the first Individual objects to be identified within a time period The time-domain distance, Indicates the first Individual objects to be identified within a time period The standard deviation of the voltage subsequence Indicates the first Individual objects to be identified within a time period The standard deviation of the voltage subsequence Indicates the time period The length of the voltage subsequence, Indicates the first Individual objects to be identified within a time period The first voltage subsequence One element, Indicates the first Individual objects to be identified within a time period The first voltage subsequence Each element.

[0041] The frequency domain distance calculation process is as follows: Perform Fourier transform on the voltage sequence of the object to be identified in different time periods to obtain the spectrum sequence in that time period; based on the spectrum sequence in each time period, calculate the absolute value of the standard deviation difference of the spectrum sequence of each object to be identified in the same time period for each pair of objects to be identified to obtain the frequency domain fluctuation difference; calculate the absolute value of the difference of each element of the spectrum sequence of each object to be identified in the same time period for each pair of objects to be identified and sum them to obtain the sum of the frequency domain element differences; sum the frequency domain fluctuation difference and the sum of the frequency domain element differences to obtain the frequency domain distance between the two objects to be identified in that time period.

[0042] The formula for calculating the frequency domain distance is: ; in, Indicates the first The and the first Individual objects to be identified within a time period Frequency domain distance, Indicates the first Individual objects to be identified within a time period The standard deviation of the upper spectral sequence, Indicates the first Individual objects to be identified within a time period The standard deviation of the upper spectral sequence, Indicates the time period The length of the upper spectral sequence, Indicates the first Individual objects to be identified within a time period The first of the upper spectral sequence One element, Indicates the first Individual objects to be identified within a time period The first of the upper spectral sequence Each element.

[0043] The target distance is calculated as follows: the product of the time domain weight and the time domain distance is used as the time domain weighted distance, and the product of the frequency domain weight and the frequency domain distance is used as the frequency domain weighted distance. The target distance is obtained by summing the time domain weighted distance and the frequency domain weighted distance for each time period.

[0044] The formula for calculating the target distance is: ; in, This represents the target distance between two objects to be identified. Indicates the number of time periods. Indicates time period The time-domain weights, Indicates the first The and the first Individual objects to be identified within a time period Frequency domain distance, Indicates time period Frequency domain weights, Indicates the first The and the first Individual objects to be identified within a time period Frequency domain distance.

[0045] Clustering is performed on all objects to be identified using the target distance as the distance metric, resulting in multiple clusters. Feasible clustering algorithms include DPC and K-means. For each cluster, the Kendall correlation coefficient is calculated between the user to be identified and each bin to be identified, and between each bin to be identified and each branch bin. The object with the highest Kendall correlation coefficient is selected as the member object, where each user belongs to a unique bin, and each bin belongs to a unique branch bin.

[0046] During peak electricity consumption periods, multiple devices start up simultaneously, resulting in rich harmonics and complex time-domain fluctuations. During off-peak periods, devices are individually stable with obvious time-domain characteristics. Directly relying on the overall Euclidean distance can smooth out changes across different periods, leading to poor clustering results. Segmented distance calculation can dynamically adapt to voltage changes during peak and off-peak periods. A single Euclidean distance can only capture changes in the time domain, resulting in poor clustering. For example, purely resistive loads like electric heaters have large time-domain fluctuations and few harmonics, while nonlinear loads like variable frequency air conditioners have stable time domains but rich harmonics. Compared to using the traditional Euclidean distance across the entire sequence as a distance metric, calculating the target distance based on time and frequency domain distances and time and frequency domain discriminative power across different time periods as a clustering distance metric allows for accurate capture of the dual differences between different objects to be identified, thanks to the two-dimensional distance in both the time and frequency domains.

[0047] Step S4: Input the obtained membership relationships of each object to be identified into the system to generate a low-voltage distribution area topology map to complete the topology identification.

[0048] The low-voltage transformer area topology identification method of the present invention calculates the target distance based on the time and frequency domain distance and time and frequency domain discrimination in different time periods as a distance metric for clustering. The time and frequency domain dual-dimensional distance can accurately capture the dual differences between different objects to be identified, thereby improving the identification accuracy of the membership relationship of the objects to be identified.

[0049] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for identifying the topology of low-voltage distribution areas based on electricity meter data, characterized in that, include: Obtain the voltage sequence of each object to be identified in the low-voltage distribution area; The voltage sequences are segmented according to time periods, and the time-domain and frequency-domain discriminant properties of each object to be identified are calculated pairwise within the same time period. The calculation process of the time-domain discriminant property includes: Based on the voltage sequence in each time period, the ratio of the Manhattan distance to the maximum Manhattan distance of each object to be identified in the same time period is calculated pairwise, and the sum of the ratios is used as the temporal distinguishability of different objects to be identified in that time period. The calculation process for the frequency domain discrimination includes: Perform Fourier transform on the voltage sequence of the object to be identified within different time periods to obtain the spectral sequence within that time period; Based on the spectral sequences within each time period, the ratio of the Manhattan distance to the maximum Manhattan distance for each object to be identified in the same time period is calculated pairwise, and the sum of the ratios is used as the frequency domain distinguishability of different objects to be identified in that time period. The time-domain distance and frequency-domain distance of the two objects to be identified are calculated in each time period. The time-domain discrimination and frequency-domain discrimination are assigned as weights to the time-domain distance and frequency-domain distance. The time-domain distance and frequency-domain distance are weighted and fused to calculate the target distance. Based on the target distance, each object to be identified is clustered. The membership relationship of each object to be identified is calculated based on the clustering results. The affiliation relationships of each object to be identified are input into the system to generate a low-voltage distribution area topology map, thus completing the topology identification.

2. The low-voltage distribution area topology identification method based on meter data according to claim 1, characterized in that, The calculation process of the time-domain distance includes: Based on the voltage sequences within each time period, the absolute value of the standard deviation difference of the voltage sequences of each object to be identified in the same time period is calculated pairwise to obtain the time-domain fluctuation difference; the absolute value of the difference of each element of the voltage sequence of each object to be identified in the same time period is calculated pairwise and summed to obtain the sum of time-domain element differences; the sum of the time-domain fluctuation difference and the sum of time-domain element differences is used as the time-domain distance between the two objects to be identified in that time period.

3. The low-voltage distribution area topology identification method based on meter data according to claim 1, characterized in that, The calculation process of the frequency domain distance includes: Perform Fourier transform on the voltage sequence of the object to be identified within different time periods to obtain the spectral sequence within that time period; Based on the spectral sequences within each time period, the absolute value of the standard deviation difference of the spectral sequences of each object to be identified in the same time period is calculated pairwise to obtain the frequency domain fluctuation difference; the absolute value of the difference of each element of the spectral sequence of each object to be identified in the same time period is calculated pairwise and summed to obtain the sum of frequency domain element differences; the sum of the frequency domain fluctuation difference and the sum of frequency domain element differences is taken as the frequency domain distance between the two objects to be identified in that time period.

4. The low-voltage distribution area topology identification method based on meter data according to claim 1, characterized in that, The target distance is calculated as follows: The time-domain and frequency-domain discrimination are normalized to obtain the time-domain weight and frequency-domain weight. The product of the time-domain weight and the time-domain distance is used as the time-domain weighted distance, and the product of the frequency-domain weight and the frequency-domain distance is used as the frequency-domain weighted distance. The target distance is obtained by summing the time-domain weighted distance and the frequency-domain weighted distance for each time period.

5. The low-voltage distribution area topology identification method based on meter data according to claim 1, characterized in that, Using the target distance as a distance metric, all objects to be identified are clustered to obtain multiple clusters. The objects to be identified include branch boxes, table boxes, and users. For a cluster, the Kendall coefficient between the user to be identified and each table box to be identified, and between the table box to be identified and each branch box to be identified, are calculated. The object to be identified with the highest correlation of the Kendall coefficient is selected as the member object. One user to be identified belongs to a unique table box to be identified, and one table box to be identified belongs to a unique branch box to be identified.

6. The low-voltage distribution area topology identification method based on meter data according to claim 1, characterized in that, When segmenting voltage sequences according to time periods, the optimal segmentation method is selected. The specific operation is as follows: The voltage sequence of each object to be identified is divided into multiple voltage subsequences. Fourier transform is used to obtain the spectral sequence of each voltage subsequence in each object to be identified. The standard deviation of the frequency amplitude of each voltage subsequence is calculated based on the spectral sequence. The variance is calculated based on the standard deviation of the frequency amplitude of each voltage subsequence. The optimal segmentation method of the voltage sequence is selected according to the variance.

Images