Phase Identification Method and Device for Distribution Transformer Area Based on Cluster Analysis

Abstract

The present invention discloses a method and device for identifying the phase of a power distribution area based on clustering analysis. The steps of the method include: S01. Obtaining two or more kinds of operation time series data of each user electric meter in the power distribution area to be identified; S02. Respectively performing comprehensive processing on each operation time series data of each user electric meter to obtain a comprehensive time series data set; S03. Using spectral clustering to perform clustering analysis on the comprehensive time series data set, and dividing it into 3 clusters to respectively correspond to the ABC three phase sequences of the phase of the power distribution area; S04. Obtaining the phase identification result of the power distribution area to be identified according to the clustering analysis result. The present invention can realize the phase identification of batch data, and has the advantages of simple implementation method, low cost, high identification accuracy and efficiency, wide single-time identification range, safety and reliability, etc.

Description

Technical Field

[0001] The present invention relates to the technical field of phase identification of a transformer substation, and in particular to a method and a device for phase identification of a transformer substation based on cluster analysis. Background Art

[0002] In low-voltage distribution networks, three-phase imbalance is a common problem in low-voltage (LV) distribution networks. The main cause is the unbalanced load on the low-voltage side. Three-phase load imbalance not only reduces the power supply efficiency of the line and distribution transformer, but in severe cases may cause damage to phase conductors, switches, distribution transformers and other devices, leading to power grid failures and potential safety hazards. For example, when the low-voltage three-phase load imbalance problem occurs, a single line may have no current or multiple currents coexist, which will increase the current carrying pressure of the line, and excessive losses will continue to increase. When the line can no longer withstand excessive current and the loss is overloaded, the line will also be quickly damaged, causing line safety problems. In addition, since there will be many electrical appliances connected to the distribution network, if the current is too large, the transformer carried inside the appliance will lose its function and cannot effectively control the current. The destruction of the overloaded line will cause the temperature of the appliance to rise rapidly, and then the transformer will trip quickly.

[0003] Although the three-phase load is balanced as much as possible when configuring each user, the load on each phase cannot be completely balanced due to the asynchronous startup time. The three-phase imbalance problem is further complicated by the frequent changes in user connections on each phase, asymmetric line configuration, and the existence of single-phase distributed generators. In order to maintain the balance of the three-phase distribution network, the first task is to identify the exact phase line where all electricity users are located, so higher technical requirements are currently put forward for phase identification.

[0004] The current phase (area) identification usually uses a phase / area identifier to identify signal crosstalk between stations. However, this method can only identify one user at a time. Not only is the identification speed slow, but there are also problems such as the need to install additional terminal equipment, high price and difficulty in carrying. It is difficult to meet the current phase identification needs of complex topology areas. Therefore, there is an urgent need for an area phase identification method suitable for batch processing, which can be fast, data-driven, have a large identification range, high security, and reduce hardware and labor costs. Summary of the invention

[0005] The technical problem to be solved by the present invention is: in response to the technical problems existing in the prior art, the present invention provides a cluster analysis-based phase identification method and device with simple implementation method, low cost, high identification accuracy and efficiency, wide single identification range, and safety and reliability.

[0006] To solve the above technical problems, the technical solution proposed by the present invention is as follows:

[0007] A method for identifying the phase of a power distribution area based on clustering analysis, the steps including:

[0008] S01. Obtain the time-series data of the operating parameters and the time-series data of the signal-to-noise ratio of each user's electricity meter in the power distribution area to be identified;

[0009] S02. Respectively perform comprehensive processing on the time-series data of the operating parameters and the time-series data of the signal-to-noise ratio of each user's electricity meter to obtain a comprehensive time-series data set;

[0010] S03. Use spectral clustering to perform clustering analysis on the comprehensive time-series data set, and divide it into 3 clusters to respectively correspond to the ABC three phase sequences of the phase of the power distribution area;

[0011] S04. Obtain the output of the phase identification result of the power distribution area to be identified according to the clustering analysis result.

[0012] Further, the time-series data of the operating parameters includes voltage time-series data and / or current time-series data.

[0013] Further, after step S01 and before step S02, it further includes first performing low-pass filtering on the obtained time-series data of the operating parameters to filter out the data contaminated by high frequencies, and then performing homomorphic filtering to filter out the data contaminated by multiplicative noise.

[0014] Further, after step S01 and before step S02, it further includes performing preprocessing and normalization on the time-series data of the operating parameters and the time-series data of the signal-to-noise ratio to filter out those exceeding a preset threshold.

[0015] Further, the comprehensive processing in step S02 is specifically weighted processing, that is, using the time-series data of the signal-to-noise ratio of each user's electricity meter as a weight coefficient to weight the operating time-series data to form comprehensive time-series data.

[0016] Further, step S03 includes:

[0017] Form multiple samples according to the comprehensive time-series data set and calculate the parameters required for spectral clustering;

[0018] Use the parameters required for calculating the spectral clustering to perform k-means clustering on each of the samples, divide the comprehensive time-series data set into 3 clusters, and obtain a clustering result.

[0019] Further, the parameters required for the spectral clustering include: adjacency matrix W, degree matrix D, non-normalized Laplacian matrix L, n-dimensional eigenvectors corresponding to the k smallest eigenvalues of the Laplacian matrix L, and a matrix M composed of k n-dimensional eigenvectors in an n×k dimension.

[0020] Further, in step S04, obtain the phase test result of the output feeder, and comprehensively combine the phase test result of the output feeder with the final recognition result obtained according to the clustering analysis result.

[0021] A phase recognition device for a power distribution area based on clustering analysis, comprising:

[0022] An acquisition module, configured to acquire the time series data of the operation parameters and the time series data of the signal-to-noise ratio of each user electric meter in the power distribution area to be recognized;

[0023] A comprehensive processing module, configured to comprehensively process the time series data of the operation parameters and the time series data of the signal-to-noise ratio of each user electric meter respectively to obtain a comprehensive time series data set;

[0024] A clustering analysis module, configured to perform clustering analysis on the comprehensive time series data set using spectral clustering, and divide it into 3 clusters to respectively correspond to the ABC three phase sequences of the power distribution area phase;

[0025] A recognition result output module, configured to output the phase recognition result of the power distribution area to be recognized according to the clustering analysis result.

[0026] A computer device, comprising a processor and a memory, the memory is used to store a computer program, and the processor is used to execute the computer program, wherein the processor is used to execute the computer program to execute the method as described above.

[0027] Compared with the prior art, the advantages of the present invention are as follows:

[0028] 1. The present invention uses spectral clustering to perform clustering analysis on the operation data of the intelligent electric meters in the power distribution area, and further fuses the noise measurement data into the voltage measurement data at the same time to indirectly improve the signal-to-noise ratio of the voltage measurement data. Then, the spectral clustering algorithm is applied to the processed data to complete the phase recognition, so that the influence of noise can be fully considered, and the measurement data containing noise dimensions can be used to more accurately characterize the phase characteristics of the power distribution area, thereby effectively improving the recognition accuracy and universality.

[0029] 2. The present invention weights the voltage operation data by using the time series data of the signal-to-noise ratio as the weight coefficient, so that after weighting, the proportion of data with a high signal-to-noise ratio is increased and the proportion of data with a low signal-to-noise ratio is reduced. The data corresponding to a high signal-to-noise ratio has higher reliability, which indirectly realizes data filtering. Therefore, the accuracy of phase recognition can be effectively improved finally, and at the same time, the complexity of data processing and the processing efficiency can be reduced.

[0030] 3. The present invention can accurately identify the phase of the user, timely correct and identify the substation area information of the users in the distribution network, so as to facilitate the accurate calculation of the line loss situation of the distribution network, etc., and effectively monitor the working conditions of electrical components in the distribution network in real time. It can not only effectively improve the ability of refined management of the power system, but also reduce resource waste and protect the environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic flowchart of the implementation process of the substation area phase identification method based on clustering analysis in this embodiment.

[0032] Figure 2 It is a schematic flowchart of the process of collecting and processing voltage time series data in this embodiment.

[0033] Figure 3 It is a schematic flowchart of the process of collecting and processing signal-to-noise ratio time series data in this embodiment.

[0034] Figure 4 It is a detailed schematic flowchart of realizing substation area phase identification based on spectral clustering in this embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] The present invention will be further described below in conjunction with the accompanying drawings of the specification and specific preferred embodiments, but the protection scope of the present invention is not limited thereby.

[0036] As Figure 1 shown, the steps of the substation area phase identification method based on clustering analysis in this embodiment include:

[0037] S01. Obtain the time series data of the operating parameters and the time series data of the signal-to-noise ratio of each user's electric meter in the substation area to be identified;

[0038] S02. Respectively perform comprehensive processing on the time series data of the operating parameters and the time series data of the signal-to-noise ratio of each user's electric meter to obtain a comprehensive time series data set;

[0039] S03. Use spectral clustering to perform clustering analysis on the comprehensive time series data set, and divide it into 3 clusters to respectively correspond to the ABC three phase sequences of the substation area phase;

[0040] S04. Obtain the output of the phase identification result of the substation area to be identified according to the clustering analysis result.

[0041] The topology of the power distribution area is complex, and there will be a huge amount of operation data of smart meters. The operation data is time-series data, that is, related to the time sequence, and there may also be a large amount of noisy data, invalid data, etc. in the data. If the conventional clustering analysis is directly performed on the huge amount of meter operation data, the clustering implementation is complex, which takes a lot of time and has low real-time performance. This embodiment fully considers the above characteristics of the phase identification in the power distribution area and adopts spectral clustering for the clustering analysis of the power distribution area phase. Spectral clustering is an algorithm evolved from graph theory. Its main idea is to regard all data as points in space, and these points can be connected by edges. The edge weight value between two points with a large distance is low, while the edge weight value between two points with a small distance is high. By cutting the graph composed of all data points, the sum of the edge weights between different subgraphs after cutting is as low as possible, and the sum of the edge weights within the subgraph is as high as possible, so as to achieve the purpose of clustering. Spectral clustering only needs the similarity matrix between data, which is very suitable for processing the clustering of sparse data. And because of the use of dimensionality reduction, the complexity can be greatly reduced when processing the clustering of high-dimensional data. This embodiment makes full use of the above characteristics of spectral clustering to realize the phase identification in the power distribution area, so as to be able to achieve batch identification at one time and greatly improve the identification effect.

[0042] However, in the actual power grid application environment, the voltage data transmitted to the user meter end through the power line is not ideal data. The voltage data contains phase information and will correspondingly contain various line noises and measurement errors, and it is difficult to directly remove such noise signals. Therefore, directly performing spectral clustering analysis on the operation data after noise superposition is likely to generate clustering errors and lead to identification errors. Especially for the specific phases in the power distribution area, namely the three phases of A, B, and C, a large amount of noise interference is very likely to cause the problem of identification errors. On the basis of using spectral clustering to perform clustering analysis on the operation data of the smart meters in the power distribution area, this embodiment further considers the dimension of the noise, comprehensively processes the voltage measurement data and the noise measurement data (signal-to-noise ratio), that is, fuses the noise measurement data into the voltage measurement data to indirectly improve the signal-to-noise ratio of the voltage measurement data, and then applies the spectral clustering algorithm to the processed data to complete the phase identification, so as to fully consider the influence of the noise and more accurately characterize the phase characteristics of the power distribution area by using the measurement data containing the noise dimension, thereby effectively improving the identification accuracy and universality.

[0043] In this embodiment, the operation parameter time-series data includes voltage time-series data, current time-series data, etc. In fact, various parameters that can characterize the phase characteristics of the power distribution area operation data can be used, and specific selection can be made according to actual needs.

[0044] In this embodiment, after step S01 and before step S02, the acquired operation parameter time series data is first subjected to low-pass filtering to filter out data polluted by high frequency, and then subjected to homomorphic filtering to filter out data polluted by multiplicative noise. In the operation data of the electric energy meter, there may be not only data polluted by high frequency, but also pollution polluted by multiplicative noise, and multiplicative noise and high frequency data are difficult to be directly filtered out at the same time. After step S01 and before step S02 of this embodiment, after the operation data of the electric energy meter is acquired, the operation parameter time series data is first subjected to low-pass filtering by a low-pass filter to filter out data polluted by high-frequency colored noise, and then the voltage sequence data after low-pass filtering is further subjected to homomorphic filtering to filter out data polluted by multiplicative noise. Considering the common interference signal frequency of the transmission line, the above-mentioned low-pass filter can specifically select a cut-off frequency of 11KHz, and considering the common multiplicative noise characteristics of the transmission line, the above-mentioned homomorphic filtering can select a Butterworth filter.

[0045] This embodiment takes into account the influence of noise and combines the characteristics of the transmission line. By combining a two-stage digital filtering method of low-pass filtering and then homomorphic filtering for noise filtering, it can effectively filter out the data in the operation data of the electric energy meter that is contaminated by noise and affects the recognition accuracy. It can further make up for the defects that exist when the spectral clustering algorithm is directly used for phase recognition of voltage data, thereby further improving the accuracy and universality of phase recognition achieved by spectral clustering.

[0046] In this embodiment, after step S01 and before step S02, the acquired operation parameter time series data is also preprocessed to filter out data exceeding a preset threshold value. For example, the data whose voltage exceeds a preset threshold value (which can be configured according to actual needs, such as 300V) can be specifically set to be filtered out to filter out erroneous measurement data that obviously exceeds the normal range.

[0047] In this embodiment, after step S01 and before step S02, the operating parameter time series data and the signal-to-noise ratio time series data are normalized to obtain normalized operating parameter time series data and signal-to-noise ratio time series data, so as to facilitate subsequent processing of the operating parameter time series data and the signal-to-noise ratio time series data in a unified dimension.

[0048] like Figure 2 As shown, in this embodiment, the voltage sequence data collected from the user's electric meter is pre-processed to remove obviously erroneous data (such as data with a voltage value exceeding 300v), and then the pre-processed data is normalized; Figure 3 As shown, in this embodiment, the collected signal-to-noise ratio sequence data of the user's electricity meter is preprocessed to remove obviously erroneous data, and then the preprocessed data is normalized.

[0049] In step S02 of this embodiment, the comprehensive processing is specifically weighted processing, that is, the signal-to-noise ratio time-series data of each user's electricity meter is used as a weight coefficient to weight the operation time-series data to form comprehensive time-series data. By using the signal-to-noise ratio time-series data as a weight coefficient for weighting, if the signal-to-noise ratio of the voltage operation data is high, the corresponding signal-to-noise ratio time-series data is higher, and the corresponding weight coefficient is higher. On the contrary, if the signal-to-noise ratio of the voltage operation data is low, the corresponding signal-to-noise ratio time-series data is lower, and the corresponding weight coefficient is lower. Then, after weighting, the data with a high signal-to-noise ratio can be increased and the data with a low signal-to-noise ratio can be reduced, that is, the proportion of the data with a high signal-to-noise ratio is increased. The data with a high signal-to-noise ratio has higher reliability, indirectly realizing data filtering processing. Therefore, the accuracy of phase recognition can be effectively improved finally. At the same time, compared with the traditional filtering processing method, the complexity and processing efficiency of data processing can be greatly reduced.

[0050] In step S02 of this embodiment, specifically, the voltage measurement data and the signal-to-noise ratio data are first weighted, and then the spectral clustering algorithm is applied to the processed data to complete phase recognition. Of course, other comprehensive data processing methods can also be considered according to actual needs. The key is to form data containing noise dimensions in the voltage measurement data to improve the accuracy of subsequent phase recognition.

[0051] In this embodiment, the specific steps of step S03 include:

[0052] Form multiple samples according to the comprehensive time-series data set and calculate the parameters required for spectral clustering;

[0053] Use the parameters required for calculating spectral clustering to perform k-means clustering on each sample, divide the comprehensive time-series data set into 3 clusters, and obtain the clustering result.

[0054] In this embodiment, the parameters required for spectral clustering include: adjacency matrix W, degree matrix D, unnormalized Laplacian matrix L, n-dimensional eigenvectors corresponding to the k smallest eigenvalues of Laplacian matrix L, and a matrix M composed of k n-dimensional eigenvectors with dimensions n×k.

[0055] In step S04 of this embodiment, specifically, the output feeder phase test result is obtained, and the output feeder phase test result and the final recognition result are obtained according to the clustering analysis result. For example, when the phase test result obtained based on the output feeder is the same as the final recognition result obtained according to the above clustering analysis result, it is confirmed that the above recognition result is valid and the final recognition result is output. If the phase test result obtained from the output feeder is different from the result obtained according to the above clustering analysis, further verification is required to ensure the accuracy of the final recognition result.

[0056] In a specific application embodiment, steps S03 and S04 specifically include: weighting the user electricity meter voltage sequence data and the signal-to-noise ratio sequence data processed in step S02, and then using the weighted voltage sequence data as the input data for the spectral clustering algorithm. According to the requirements of the spectral clustering algorithm, preprocess the input data, including calculating the adjacency matrix W and the degree matrix D, calculating the unnormalized Laplacian matrix L, calculating the n-dimensional eigenvectors corresponding to the k smallest eigenvalues of matrix L, forming an n×k-dimensional matrix M with the k n-dimensional eigenvectors, row-normalizing matrix M, and each row represents a sample. Then perform the k-means clustering algorithm on the n samples to obtain the clustering result, and use the normalized spectral clustering to divide the data set into 3 clusters (corresponding to phases A, B, and C). Finally, combine the output feeder phase measurement results to output the phase recognition result based on the voltage acquisition data. The phase of the transformer area corresponds to the three phase sequences A, B, and C, and there are specific angular relationships between the phase sequences.

[0057] Taking the realization of transformer area phase recognition based on clustering analysis in a specific application embodiment as an example, the above method of the present invention will be further described as follows. As Figure 4 shown, the detailed steps for realizing transformer area phase recognition based on clustering analysis in this embodiment are as follows:

[0058] Step S1: Data acquisition and preprocessing

[0059] Step S101: Preprocess the collected voltage sequence data of the user electricity meter, eliminate the obviously incorrect data (such as data with voltage values exceeding 300V), and then normalize the preprocessed data.

[0060] Step S102: Preprocess the collected signal-to-noise ratio sequence data of the user electricity meter, eliminate the obviously incorrect data, and then normalize the preprocessed data.

[0061] Step S2: Weight the processed user electricity meter voltage sequence data and signal-to-noise ratio sequence data in step S1;

[0062] Step S3: Use the weighted voltage sequence data as the input data for the spectral clustering algorithm. According to the requirements of the spectral clustering algorithm, preprocess the input data, including calculating the adjacency matrix W and the degree matrix D, calculating the unnormalized Laplacian matrix L, calculating the n-dimensional eigenvectors corresponding to the k smallest eigenvalues of matrix L, forming an n×k-dimensional matrix M with the k n-dimensional eigenvectors, row-normalizing matrix M, and each row represents a sample. Then perform the k-means clustering algorithm on the n samples to obtain the clustering result, and use the normalized spectral clustering to divide the data set into 3 clusters (corresponding to phases A, B, and C);

[0063] Step S4. Based on the clustering result obtained in Step S3 and combined with the output feeder phase measurement result, output the phase recognition result based on the voltage acquisition data.

[0064] The above-mentioned phase recognition method of the present invention can accurately recognize the phase of users, timely correct and identify the substation area information of users in the distribution network, so as to facilitate accurate calculation of the line loss situation of the distribution network, etc., and effectively monitor the working conditions of electrical components in the distribution network in real time. It can not only effectively improve the ability of refined management of the power system, but also reduce resource waste and protect the environment.

[0065] The substation area phase recognition device based on clustering analysis in this embodiment includes:

[0066] An acquisition module, configured to acquire the operation parameter time series data and the signal-to-noise ratio time series data of each user electric meter in the substation area to be recognized;

[0067] A comprehensive processing module, configured to comprehensively process the operation parameter time series data and the signal-to-noise ratio time series data of each user electric meter respectively to obtain a comprehensive time series data set;

[0068] A clustering analysis module, configured to perform clustering analysis on the comprehensive time series data set using spectral clustering, and divide it into 3 clusters to respectively correspond to the ABC three phase sequences of the substation area phase;

[0069] An identification result output module, configured to output the phase recognition result of the substation area to be recognized according to the clustering analysis result.

[0070] In this embodiment, a preprocessing module is further provided between the acquisition module and the comprehensive processing module. The preprocessing module includes a filtering unit, configured to preprocess the acquired operation parameter time series data to filter out data exceeding a preset threshold.

[0071] The preprocessing module further includes a normalization unit, configured to perform normalization processing on the operation parameter time series data and the signal-to-noise ratio time series data.

[0072] In this embodiment, the comprehensive processing in the comprehensive processing module is specifically weighted processing, that is, using the signal-to-noise ratio time series data of each user electric meter as a weight coefficient to weight the operation time series data to form comprehensive time series data.

[0073] In this embodiment, the clustering analysis module includes:

[0074] Form multiple samples according to the comprehensive time series data set and calculate the parameters required for spectral clustering, including the adjacency matrix W, the degree matrix D, the unnormalized Laplacian matrix L, the n-dimensional eigenvectors corresponding to the k smallest eigenvalues of the Laplacian matrix L, the n×k-dimensional matrix M composed of k n-dimensional eigenvectors, etc.;

[0075] Perform k-means clustering on each of the samples using the parameters required for computing spectral clustering, divide the comprehensive time-series data set into 3 clusters, and obtain the clustering result.

[0076] In this embodiment, the recognition result output module specifically obtains the output feeder phase test result, and synthesizes the output feeder phase test result and the final recognition result obtained according to the clustering analysis result.

[0077] This embodiment corresponds one-to-one with the above-mentioned substation area phase recognition method based on clustering analysis, and will not be elaborated here one by one.

[0078] This embodiment also provides a computer device, including a processor and a memory. The memory is used to store a computer program, and the processor is used to execute the computer program to execute the above-mentioned substation area phase recognition method based on clustering analysis.

[0079] The above are only the preferred embodiments of the present invention, and do not impose any form of limitation on the present invention. Although the present invention has been disclosed above with the preferred embodiments, it is not intended to limit the present invention. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of the protection of the technical solution of the present invention.

Claims

1. A method for identifying the phase of a power distribution area based on clustering analysis, characterized in that the steps Including: S01. Obtain the time - series data of the operating parameters and the time - series data of the signal - to - noise ratio of each user's electricity meter in the sub - area to be identified; S02. Respectively perform comprehensive processing on the time - series data of the operating parameters and the time - series data of the signal - to - noise ratio of each user's electricity meter to obtain a comprehensive time - series data set; The comprehensive processing is specifically weighted processing, that is, using the time - series data of the signal - to - noise ratio of each user's electricity meter as a weight coefficient to weight the time - series data of the operating parameters to form comprehensive time - series data; S03. Use spectral clustering to perform clustering analysis on the comprehensive time - series data set, and divide it into 3 clusters to respectively correspond to the ABC three phase sequences of the sub - area phase; S04. Obtain the phase identification result output of the sub - area to be identified according to the clustering analysis result; After step S01 and before step S02, it also includes first performing low - pass filtering on the obtained time - series data of the operating parameters to filter out the data contaminated by high - frequency, and then performing homomorphic filtering to filter out the data contaminated by multiplicative noise.

2. The method for identifying the phase of a power distribution area based on clustering analysis according to claim 1, wherein The time - series data of the operating parameters includes voltage time - series data and / or current time - series data.

3. The method for identifying the phase of a transformer substation area based on clustering analysis according to claim 1, wherein After step S01 and before step S02, it also includes pre - processing and normalization processing of filtering out the data exceeding the preset threshold for the time - series data of the operating parameters and the time - series data of the signal - to - noise ratio.

4. The method for identifying the phase of a transformer substation area based on clustering analysis according to any one of claims 1 to 3, characterized in that Step S03 includes: Forming multiple samples according to the comprehensive time - series data set and calculating the parameters required for spectral clustering; Using the parameters required for calculating the spectral clustering to perform k - means clustering on each of the samples, dividing the comprehensive time - series data set into 3 clusters to obtain a clustering result.

5. The method for identifying the phase of a transformer substation area based on clustering analysis according to claim 4, wherein The parameters required for the spectral clustering include: adjacency matrix W, degree matrix D, unnormalized Laplacian matrix L, n - dimensional eigenvectors corresponding to the k smallest eigenvalues of the Laplacian matrix L, and a matrix M composed of k n - dimensional eigenvectors with n×k dimensions.

6. The method for identifying the phase of a transformer substation area based on cluster analysis according to any one of claims 1 to 3, characterized in that In step S04, obtain the output feeder phase test result, and synthesize the output feeder phase test result and the final identification result obtained according to the clustering analysis result.

7. A phase identification device for a power distribution area based on clustering analysis, characterized in that, Including: An acquisition module for obtaining the time - series data of the operating parameters and the time - series data of the signal - to - noise ratio of each user's electricity meter in the sub - area to be identified; A comprehensive processing module for respectively performing comprehensive processing on the time - series data of the operating parameters and the time - series data of the signal - to - noise ratio of each user's electricity meter to obtain a comprehensive time - series data set; The comprehensive processing is specifically weighted processing, that is, using the time - series data of the signal - to - noise ratio of each user's electricity meter as a weight coefficient to weight the time - series data of the operating parameters to form comprehensive time - series data; A clustering analysis module for using spectral clustering to perform clustering analysis on the comprehensive time - series data set, and dividing it into 3 clusters to respectively correspond to the ABC three phase sequences of the sub - area phase; An identification result output module for obtaining the phase identification result output of the sub - area to be identified according to the clustering analysis result; After the acquisition module and before the comprehensive processing module, there is also a filtering module for first performing low - pass filtering on the obtained time - series data of the operating parameters to filter out the data contaminated by high - frequency, and then performing homomorphic filtering to filter out the data contaminated by multiplicative noise.

8. A computer device, comprising a processor and a memory, the memory being used for storing a computer program, and the processor being used for executing the computer program, characterized in that The processor is used to execute the computer program to execute the method described in any one of claims 1 - 6.

Images