Zone area voltage multi-dimensional feature portrait construction method based on fusion ISSA clustering

Abstract

The invention relates to a transformer area voltage multi-dimensional feature portrait construction method based on fusion ISSA clustering. The method comprises the following steps: S1, acquiring sampling point voltage and load data of each user in a transformer area; step S2, based on the voltage and load data of the sampling point, calculating a voltage feature tag considering security, stability and economic dimensions; and S3, determining an optimal clustering cluster center combination by using a K-means clustering algorithm based on ISSA and taking the voltage feature tag as a clustering object, and distributing the voltage feature tag into a cluster to obtain a clustering result, thereby realizing construction of the transformer area voltage multi-dimensional feature portrait. Compared with the prior art, the method achieves the optimal determination of the clustering center, fully considers the multi-dimensional factors of the transformer area and the difference between the users in the transformer area, and effectively avoids the falling into a local optimal solution.

Description

Technical Field

[0001] The present invention relates to the technical field of electricity consumption big data, and in particular to a method for constructing a multi-dimensional feature portrait of a transformer substation voltage based on fusion ISSA (Salp Swarm Algorithm, SSA) clustering. Background Art

[0002] In recent years, with the installation and application of more and more intelligent sensor devices in power systems, power systems can obtain power data from smart meters and digital protection devices at all times. An important research topic in the current power system field is how to make good use of the collected power big data. Data mining technology is considered to be the most commonly used data processing method for increasing amounts and dimensions of data in current domestic and international related research fields.

[0003] At present, there are many feature extraction studies based on data mining and other methods to obtain substation feature labels. Some scholars have proposed studies on segmentation areas and segmentation targets for power customers based on segmentation targets. For traditional users, it is necessary to focus on user value research and build a user value evaluation system. User portraits can categorize data to help managers quickly understand user characteristics and formulate corresponding management methods, which has been verified in practice in many fields.

[0004] At present, portrait analysis research is often used to focus on the analysis of target power consumption characteristics, but there are relatively few studies on the voltage characteristics analysis of low-voltage substations based on clustering. In addition, how to determine the optimal cluster center in the clustering process is also an urgent problem to be solved. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a method for constructing a multi-dimensional feature portrait of substation voltage based on fusion ISSA clustering, so as to solve or partially solve the problem that the clustering center is not optimal in the clustering process, which affects the clustering results.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] One aspect of the present invention provides a method for constructing a multi-dimensional feature portrait of a transformer substation voltage based on ISSA clustering fusion, comprising the following steps:

[0008] Step S1, obtaining the voltage and load data of the sampling points of each user in the substation area;

[0009] Step S2, based on the voltage and load data at the sampling point, calculating a voltage feature label considering the dimensions of safety, stability and economy;

[0010] Step S3, using the ISSA-based K-means clustering algorithm, taking the voltage feature labels as clustering objects, determining the optimal cluster center combination, assigning the voltage feature labels to the clusters, obtaining clustering results, and realizing the construction of a multi-dimensional feature portrait of the substation voltage.

[0011] As a preferred technical solution, in step S2, the safety dimension indicators include voltage deviation, voltage qualification rate and three-phase voltage imbalance; the stability dimension indicators include voltage fluctuation rate, load fluctuation rate, user voltage evaluation and user maximum voltage drop ratio; the economic dimension indicators include total electricity consumption and load rate.

[0012] As a preferred technical solution, the process of obtaining the user voltage evaluation includes the following steps:

[0013] Step S201, constructing a standardized indicator matrix of the voltage deviation size, voltage fluctuation rate and power supply distance of users in the substation area, obtaining a probability matrix by calculating the indicator proportion, calculating the information entropy and information utility value of each indicator based on the probability matrix, and normalizing to obtain the entropy weight of the indicator;

[0014] Step S202 , performing forward processing on the standardized indicator matrix to obtain a standardized extremely large indicator matrix, and obtaining a user voltage evaluation score based on the extremely large indicator matrix and the entropy weight.

[0015] As a preferred technical solution, the user voltage evaluation is calculated using the following formula:

[0016]

[0017] Among them, Y4 is the user voltage evaluation score, D i + is the distance between the ith index and the maximum value, D i - is the distance between the i-th index and the minimum value, Q Mj + is the maximum value in the standardized extremely large indicator matrix; Q Mj - is the minimum value in the standardized extremely large indicator matrix, ω j is the entropy weight of the jth indicator, and m is the number of indicators.

[0018] As a preferred technical solution, step S3 includes the following sub-steps:

[0019] Step S301, using ISSA to initialize the center positions of candidate clusters and the number of clusters;

[0020] Step S302: for each cluster, using K-means algorithm to assign voltage feature labels to the nearest cluster center;

[0021] Step S303, calculating the fitness function value based on the cluster silhouette coefficient;

[0022] Step S304, using the ISSA model with nonlinear operators and linear weights, iteratively obtains a new cluster center position, and repeats steps S302-S303 until the fitness function value satisfies the termination adjustment.

[0023] As a preferred technical solution, the ISSA model is:

[0024]

[0025] The dynamic update process of the follower's position is:

[0026]

[0027] Among them, t is the current iteration number, T max is the maximum number of iterations, x 1 j For the leader position, F j is the target position, ub j lb j are the upper and lower limits of the exploration space, c1 is the change factor, c2 and c3 are random numbers between [0,1], α(t) is a nonlinear operator, and x i,j is the follower position; x i-1,j is the neighbor follower position and β is the linear weight.

[0028] As a preferred technical solution, the following steps are also included:

[0029] Step S4, taking the clustering results as samples, inputting them into a pre-built BP neural network for iteration, and using the iterated BP neural network to identify the voltage operation mode of the substation and the type of voltage problem existing in the substation with abnormal power grid operation.

[0030] As a preferred technical solution, in the step S1, the voltage and load data of the sampling points are the voltage and load data of 96 sampling points per day. After obtaining the voltage and load data of the sampling points, the following steps are further included:

[0031] The voltage and load data of the sampling points are subjected to missing value and abnormal value processing, data transformation, data integration and data reduction processing.

[0032] Another aspect of the present invention provides an electronic device, comprising: one or more processors and a memory, wherein the memory stores one or more programs, and the one or more programs include instructions for executing the aforementioned method for constructing a multi-dimensional feature portrait of substation voltage based on fused ISSA clustering.

[0033] Another aspect of the present invention provides a computer-readable storage medium, including one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the aforementioned method for constructing a multi-dimensional feature portrait of substation voltage based on fused ISSA clustering.

[0034] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0035] (1) Optimal determination of cluster centers during clustering: The present invention utilizes a K-means clustering algorithm based on ISSA, takes the voltage feature labels as clustering objects, determines the optimal cluster center combination, distributes the voltage feature labels into clusters, obtains clustering results, and realizes the construction of a multi-dimensional feature portrait of the substation voltage. By ensuring the optimal location of the cluster center, the clustering effect is improved and the accuracy of the constructed portrait is guaranteed.

[0036] (2) Fully consider the multi-dimensional factors of the substation: Based on the voltage and load data of the sampling points, the present invention calculates the voltage feature labels that take into account the safety, stability and economy dimensions, prepares for subsequent clustering, and fully considers the indicators related to the substation voltage characteristics to achieve a full characterization of the substation voltage characteristics.

[0037] (3) Fully consider the differences among users in the substation area: The present invention constructs an indicator matrix and calculates the entropy weight of the indicator and the standardized extremely large indicator matrix respectively, so as to comprehensively and objectively describe the various indicators of the voltage of the users in the substation area.

[0038] (4) Avoiding falling into the global optimum: The present invention utilizes the ISSA model with nonlinear operators and linear weights to iteratively obtain new cluster center positions, effectively reducing the possibility of falling into the local optimum.

[0039] (5) Fully consider the operation modes of various voltage indicators in the substation in the short term in the future: The present invention uses the ISSA-based BPNN time series prediction algorithm and takes the voltage feature label as the prediction object to obtain the prediction results of the substation voltage anomaly category, thereby realizing the identification of the substation voltage operation mode and the identification of the voltage problem types existing in the substation with abnormal power grid operation, thereby improving the timeliness of substation fault management. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a flow chart of a method for constructing a multi-dimensional feature portrait of a substation voltage based on ISSA clustering fusion in an embodiment;

[0041] Figure 2 Schematic diagram of the SSE label model of the voltage characteristics of the substation in the embodiment;

[0042] Figure 3It is the topological diagram of the substation area in the embodiment;

[0043] Figure 4 It is a flow chart of ISSA-Kmeans-BPNN algorithm in the embodiment;

[0044] Figure 5 It is a comparison diagram of the convergence curves in the embodiments;

[0045] Figure 6 This is a schematic diagram for comparing the prediction results and error values ​​in the embodiment;

[0046] Figure 7 This is a schematic diagram of the prediction result of time series feature label clustering in the embodiment;

[0047] Figure 8 It is a schematic diagram of the portrait of the No. 1 station area in the embodiment. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0049] Example 1

[0050] In view of the problems existing in the above-mentioned prior art, this embodiment provides a method for constructing a multi-dimensional feature portrait of the substation voltage based on ISSA clustering. First, feature labels are extracted based on the voltage and load data of 96 sampling points every day, which is different from the existing labels to obtain a more comprehensive and specific label system. The preprocessing of the original data of the substation and feature extraction based on data mining are the first link of this study. The substation data processed by data cleaning, change, integration and other methods are used to extract multi-dimensional voltage characteristic labels through data mining; secondly, according to the constructed label system, the improved Salp Swarm Algorithm (SSA) is used to integrate the K-means clustering algorithm to perform cluster analysis on each label, and then the clustering results are used for ISSA-BPNN time series prediction, and the substation voltage portrait is constructed according to the clustering prediction results. Based on the portrait information, it can help to quickly understand the substation voltage mode and accurately perform relevant voltage management; finally, feature mining and portrait construction case practice are carried out on the voltage data of the substation in Shanghai.

[0051] See also Figure 1 , this method mainly comprises the following steps:

[0052] Step S1, obtaining the voltage and load data of the sampling points of each user in the substation area.

[0053] Specifically, this step obtains voltage and load data of 96 sampling points every day and performs data preprocessing.

[0054] Data preprocessing uses missing value and outlier processing, data transformation, data integration, data reduction and other means to clean and normalize the data, so as to ensure that the processed data has better results when mining tasks or model building, and is more adaptable to experimental needs. It not only improves the reliability, completeness and accuracy of the data, but also improves the efficiency and accuracy of the research.

[0055] Step S2, based on the voltage and load data at the sampling point, calculate the voltage feature label considering the safety, stability and economy dimensions.

[0056] This step uses direct acquisition, statistical analysis, data mining and other methods to extract voltage feature labels from the preprocessed data to evaluate voltage characteristics from different dimensions. Labels, with their semantic and short text features, have become an excellent carrier for expressing certain representation information, and labels basically do not require redundant data analysis and processing, which is more conducive to the use of label information.

[0057] Traditional methods only analyze the most basic statistical labels such as maximum, minimum, and mean, and cannot fully describe the characteristics of each dimension of the transformer area voltage. To address this problem, this step combines data mining, data dimensionality reduction, and other means to build a more comprehensive and complete labeling system.

[0058] Specifically, this step is based on the consideration of the voltage and power safety and economic dimensions of the substation, and constructs an SSE label model covering the three dimensions of safety (S), stability (S) and economy (E). Figure 2 As shown. Among them, safety refers to the fact that when the power grid is operating normally, the extracted indicators such as the voltage deviation and voltage qualification rate of the substation are based on the national standard as the safety requirement standard, and governance measures are taken for the substations in the unsafe or critical range; the labels included in stability are all obtained by mining 96 points of data in one day, considering the changes of various indicators under continuous time conditions, adopting an improved clustering algorithm, and classifying and analyzing the governance of indicators; economic efficiency considers the economic contribution value of the substation operation, and combined with the actual situation, the governance method and degree can be considered.

[0059] The calculation processes of three-phase voltage unbalance, load fluctuation rate, user voltage evaluation and user maximum voltage drop ratio are explained below.

[0060] (1) Three-phase voltage imbalance

[0061] Three-phase voltage imbalance is an important indicator for evaluating the voltage status of the power grid in the power system. It reflects the degree of imbalance of the three-phase voltage in the power grid. The balance of three-phase voltage is one of the important factors affecting the service life of power equipment and the stable operation of the power grid. Therefore, ensuring the balance of three-phase voltage in the power grid and accurately calculating the imbalance of three-phase voltage are crucial for the safe and stable operation of the power system. Its expression is:

[0062]

[0063] Where, Y1 is the comprehensive unbalance of three-phase voltage; U A is the A phase voltage at any sampling point; U N is the mean value of the three-phase voltage at the same sampling point; Y2 is the maximum unbalance of the three-phase voltage in the substation; U Arms is the RMS value of phase A voltage; U avg The mean RMS value of the three-phase voltage.

[0064] (2) Load fluctuation rate.

[0065] Substation load fluctuation refers to the change in load of a specific substation in the power system over a period of time. The magnitude of this change can be reflected by the load fluctuation rate. Load fluctuations may be cyclical, seasonal or sudden, and the reasons may be weather changes, changes in people's electricity consumption habits, equipment failures and other factors. The load fluctuation rate expression is:

[0066]

[0067] Where, Y3 is the load fluctuation rate; L i is the load value of any sampling point; L avg is the load mean.

[0068] The load fluctuation rate of the substation is classified and analyzed by clustering. Load fluctuation will have a certain impact on the stability of the power system operation. When the load fluctuation is large, it may cause voltage fluctuations, frequency deviations, equipment overloads and other problems, and even cause instability or failure of the power system.

[0069] (3) User voltage evaluation.

[0070] For all users in the substation area, based on the consideration of the differences between different users, the entropy weight method is used to calculate the information entropy of discretized indicators, and the entropy weight is used to replace the weight of the original indicator to obtain a more objective indicator weight; at the same time, the TOPSIS method is used to evaluate the voltage deviation size, voltage fluctuation rate, power supply distance and voltage qualification rate labels of users in the substation area, and finally the model evaluation score is used as the user voltage evaluation label.

[0071] First, the entropy weight method is used to determine the weights. ij] n·m Perform normalization so that there are no negative numbers in the matrix. The normalized matrix Q = [q ij ] n·m for:

[0072]

[0073] Calculate the weight of each indicator and get the probability matrix R = [r ij ] n·m , each element r in the matrix ij The calculation formula is:

[0074]

[0075] According to the probability matrix R, the information entropy and utility value of each indicator are calculated, and the entropy weight of each indicator is obtained after normalization. The calculation formula is:

[0076]

[0077] In the formula, e j is the indicator information entropy; d j is the information utility value; ω j is the index entropy weight.

[0078] Secondly, the TOPSIS model is used to evaluate the score. Similarly, for the indicator matrix P = [p ij ] n·m Perform positive processing to transform all indicators into extremely large indicators. The extremely large indicator matrix P M for:

[0079]

[0080] For the matrix P M Using the same standardization method, we get the standardized extremely large indicator matrix Q M =[q Mij ] n·m

[0081] The entropy weight ω obtained by combining the entropy weight method j Calculate the score, the evaluation formula is:

[0082]

[0083] Where, Y4 is the evaluation score; D i + is the distance between the ith index and the maximum value; D i - is the distance between the i-th index and the minimum value; Q Mj + is the maximum value in the indicator matrix; Q Mj- is the minimum value.

[0084] The final evaluation score is the defined user voltage evaluation. This label can comprehensively and objectively describe the various indicators of the user voltage in the substation area. Figure 3 Take the topological diagram of a certain area as an example.

[0085] Single-phase users use voltage deviation as the voltage offset indicator, and three-phase users use directional voltage deviation and three-phase voltage imbalance as voltage offset indicators. The higher the user voltage evaluation score label, the better the user voltage quality, the more stable the voltage, and the less affected by factors such as voltage fluctuations.

[0086] (4) Maximum pressure drop ratio of the user.

[0087] User voltage drop refers to the phenomenon that the user-side voltage in the substation area is lower than the system nominal voltage due to factors such as line resistance and voltage loss. Excessive voltage drop may affect the operation of the equipment and reduce the life of the equipment, or threaten the safety of the user, which is undoubtedly a major safety hazard. Therefore, controlling user voltage drop is of great significance to ensure the safe and efficient operation of electrical equipment, improve the quality of power supply, and maintain the stability and economic operation of the power grid. The expression is:

[0088]

[0089] Y5=[y i1 ,y i2 ,...,y im ] (14)

[0090] In the formula, y i Indicates the maximum voltage drop ratio of any station area; u i Indicates the voltage at any user side in the substation area; u o Indicates the transformer terminal voltage; Y5 indicates the maximum voltage drop ratio curve of the transformer area.

[0091] Step S3, using the ISSA-based K-means clustering algorithm, taking the voltage feature labels as clustering objects, determining the optimal cluster center combination, assigning the voltage feature labels to the clusters, obtaining the clustering results, and realizing the construction of a multi-dimensional feature portrait of the substation voltage.

[0092] This step uses the improved Salp Squid algorithm to integrate and improve the K-means clustering algorithm and the BP neural network, and performs clustering prediction analysis on the constructed substation voltage multi-dimensional feature label system. By utilizing ISSA's excellent global search capabilities, the defects of the traditional K-means algorithm, such as being prone to falling into local optimal solutions, relying on random initial center selection, and redundant distance calculation, are solved. The clustering efficiency and clustering quality are improved, making the clustering results more objective and reliable; at the same time, the prediction accuracy of the BP neural network model is improved. For indicators that are not required by national standards, the clustering results can be used as safety boundaries and compared with the prediction results for safety judgment.

[0093] (1) Improved Salp Swarm Algorithm (ISSA).

[0094] 1) Improve the leader position update model

[0095] The front end of the salp group is regarded as the group leader, and the rest are regarded as followers. The traditional algorithm position update only performs global or local search selection through c1, which is easy to fall into the local optimal solution. In view of the above problems, the algorithm of this embodiment introduces a nonlinear operator α(t) to avoid the algorithm from falling into the local optimal solution and increase the global search capability. The improved model expression is:

[0096]

[0097] Where t is the number of iterations; T max is the maximum number of iterations; x 1 j For the leader position; F j is the target position; ub j lb j is the upper and lower limits of the exploration space; c1 is the change factor; c2 and c3 are random numbers between [0,1].

[0098] Considering that when the iteration number t is small, the model search space is large, the leader direction and step size are random, and it is easy to fall into the local optimum; and as the number of iterations increases, the space becomes smaller, and the search accuracy needs to be improved to find the optimal solution. The α(t) operator is selected for constraint control, so that the leader position model can better realize global and local search.

[0099] 2) Followers’ locations are updated dynamically.

[0100] Considering that the influence of the leader on the followers is uncontrollable, in order to increase the influence of the leader on the followers and pursue the best fitness, this step adds a dynamic update mechanism to the follower position update model, introduces linear weight β, and appropriately adjusts the weight ratio according to the fitness between the position before iteration and the adjacent position, and follows the direction of the best fitness to achieve the optimal solution. The improved expression is:

[0101]

[0102] In the formula, x i,j is the follower position; x i-1,j For adjacent follower positions.

[0103] 3) Cauchy mutation.

[0104] Cauchy mutation can generate new individuals by randomly perturbing the leader in each direction, and use its perturbation characteristics and large perturbation range to increase population diversity and effectively reduce the risk of falling into a local optimal solution. The expression is:

[0105]

[0106] Where f(x) is the Cauchy distribution function; X best (t) is the global optimal solution; C(0,1) is the standard Cauchy distribution.

[0107] (2) K-means clustering algorithm based on ISSA.

[0108] The extracted voltage feature labels are used as clustering objects, and the K-means clustering algorithm integrated with the improved Salp Swarm algorithm is used for clustering analysis. The clustering results are integrated with the ISSA BP neural network time series prediction to establish the substation voltage profile. The ISSA-Kmeans-BPNN algorithm flow chart is shown in the figure below: Figure 4 shown.

[0109] The determination of the initial cluster center and the number of clusters has a significant impact on the clustering effect. In this step, K-means is integrated with ISSA to optimize the selection of the initial cluster center to avoid falling into the local optimal solution.

[0110] ISSA-K-means first uses ISSA initialization to search for candidate cluster center positions, selects cluster data cluster centers and the number of clusters K; secondly, for each cluster, the K-means algorithm is used to assign each data to the nearest cluster center. In order to judge the clustering effect, the fitness function value is determined by calculating the silhouette coefficient Si of each cluster. The closer the coefficient is to 1, the better the effect. The expression is:

[0111]

[0112] Where a is the average distance between a sample and other samples in the cluster; b is the average distance between a sample and samples in other clusters.

[0113] Then, the search mechanism of the ISSA algorithm is used to replace the original center with the updated cluster center. After several iterations, the cluster center data allocation and fitness function value calculation are repeated until the fitness function value no longer changes significantly, at which time the optimal cluster center is reached. Finally, the optimal cluster center combination is determined, and K-means is used to allocate data to various clusters to obtain the final clustering result.

[0114] Step S4, taking the clustering results as samples, inputting them into the pre-built BP neural network for iteration, and using the iterated BP neural network to realize the recognition of the voltage operation mode of the substation area, and the recognition of the voltage problem type existing in the substation area with abnormal power grid operation.

[0115] The clustering results are used as samples to input into the ISSA-BPNN prediction algorithm, and the samples are taken for forward propagation, the input and output of each layer of neurons are calculated, and the error is calculated, and the weight and threshold of the error-adjusted neural network are returned. After several iterations, the output samples that meet the error accuracy are obtained.

[0116] Evaluation index selection determination coefficient R 2 , mean absolute percentage error (MAPE) and root mean square error (RMSE) to judge the reliability of the prediction model. 2 The closer it is to 1, the closer MAPE and RMSE are to 0, which proves that the prediction effect is better. The indicator expression is:

[0117]

[0118] In the formula, y i is the true value; if is the predicted value; iavg is the true mean; n is the number of samples.

[0119] In order to verify the effectiveness of this method, the ISSA-Kmeans algorithm is used to improve the performance of the experimental data. The ISSA algorithm of this embodiment is compared with the traditional SSA, K-means, and K-means++ algorithms for convergence effect. The silhouette coefficient is used as the clustering comparison index. The convergence curve comparison diagram is shown in FIG. Figure 5 As shown, the convergence optimization effect of this embodiment is better.

[0120] To verify the effect of ISSA-BP neural network time series prediction algorithm, the voltage deviation label is selected as the experimental object, and the training set and the test set are divided into 7:3 ratios to obtain the final prediction results and error values. Figure 6 As shown, (a) BPNN prediction results, (b) SSA-BPNN prediction results, (c) ISSA-BPNN prediction results

[0121] It can be seen from the prediction results that after the ISSA-BPNN improvement, the predicted value is basically consistent with the true value, has the same fluctuation turning sampling point, and the effect is better than the traditional BPNN and SSA-BANN algorithms. At the same time, the improvement effect is judged by comparing the evaluation indicators as shown in Table 1.

[0122] Table 1 Comparison of evaluation indicators

[0123]

[0124] From Table 1, we can see that the selection of ISSA fusion BP neural network can not only improve the prediction accuracy but also improve the generalization ability of the model. In summary, it can be determined that the selection of ISSA-Kmeans-BP can indeed achieve high-efficiency and high-quality clustering and prediction of the station area data.

[0125] The following is a practical example to illustrate this method.

[0126] (1) Feature label image.

[0127] The voltage-related data of the substation area consists of 96 points of voltage and power data. Through these two parts of data, the voltage operation mode of the substation area can be understood and analyzed. The 96-point voltage and power data of the substation area are selected as the research object, and data mining and clustering prediction algorithms are used to extract feature labels and build portraits. The 96-point power data is collected every 15 minutes for the voltage and power changes within 24 hours a day to characterize the voltage characteristics in a shorter period of time.

[0128] Construct a multi-dimensional feature image of the distribution station area voltage, and the time series feature labels are as follows Figure 7 As shown, (a) voltage deviation clustering prediction result, (b) three-phase power supply imbalance clustering prediction result, (c) maximum voltage drop ratio clustering prediction result.

[0129] The non-timing indicators are shown in Table 2.

[0130] Table 2 Non-sequential label images

[0131]

[0132] For users with the first type of characteristic labels, all voltage-related indicators are within the safe range. For users with the second and third types of labels, corresponding governance measures will be implemented.

[0133] (2) User voltage profile.

[0134] Through the above-mentioned clustering results of various indicators, we can obtain multi-dimensional substation voltage characteristics, based on which we can build the final substation voltage portrait system. Taking substation 1 as an example, the comprehensive portrait results of the substation voltage label are as follows: Figure 8As shown. The portrait results clearly describe the multi-dimensional voltage situation of the substation. Substation 1 belongs to the second type of voltage deviation category, is in the safe range, has high voltage stability, and the daily voltage is qualified. It belongs to the first type of three-phase voltage imbalance clustering and is also in the safe range, so there is no need for corresponding voltage management. At the same time, it has low power consumption, medium load rate and load fluctuation rate. According to the actual situation of the users in the substation, targeted peak-shaving and valley-filling strategies can be appropriately implemented to reduce the operating pressure of the power system and improve the stability and operating efficiency of the power grid. The voltage quality of each user in the substation is high, the voltage drop is relatively small, there is no abnormal voltage situation, and no management is required.

[0135] In summary, this method provides a method for constructing a voltage SSE label system, which mines voltage characteristics from three dimensions, and obtains a substation voltage portrait according to the clustering prediction analysis of the label system. The constructed label system is clustered by K-means clustering fusion improved SSA to improve clustering accuracy, avoid falling into local optimal solutions, make clustering results more realistic, and achieve better results; the clustering results are then used for time series prediction through the BP neural network fused with ISSA. The substation voltage portrait constructed according to this method based on the prediction results can accurately and quickly identify whether the voltage operation mode of different types of substations is normal, and the types of voltage problems in the substations with abnormal power grid operation, so as to achieve more targeted voltage governance and user load operation strategies of "one substation, one plan", and effectively assist power system managers to adopt governance plans and supervise the real-time normal operation of the substation. The substation portrait is constructed using actual substation data as an example to verify the effect of this construction method, providing an important basis for voltage supervision.

[0136] Example 2

[0137] This embodiment provides an electronic device, including: one or more processors and a memory, wherein the memory stores one or more programs, and the one or more programs include instructions for executing the method for constructing a multi-dimensional feature portrait of substation voltage based on fused ISSA clustering as described in Example 1.

[0138] Example 3

[0139] This embodiment provides a computer-readable storage medium, including one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the method for constructing a multi-dimensional feature portrait of substation voltage based on fused ISSA clustering as described in Example 1.

[0140] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for constructing a multi-dimensional feature portrait of a substation voltage based on ISSA clustering, characterized in that: The steps include: Step S1, obtaining the voltage and load data of the sampling points of each user in the substation area; Step S2, based on the voltage and load data at the sampling point, calculating a voltage feature label considering the dimensions of safety, stability and economy; Step S3, using the ISSA-based K-means clustering algorithm, taking the voltage feature labels as clustering objects, determining the optimal cluster center combination, assigning the voltage feature labels to the clusters, obtaining clustering results, and realizing the construction of a multi-dimensional feature portrait of the substation voltage.

2. According to claim 1, a method for constructing a multi-dimensional feature portrait of a substation voltage based on ISSA clustering fusion, characterized in that: In the step S2, the safety dimension indicators include voltage deviation, voltage qualification rate and three-phase voltage imbalance; the stability dimension indicators include voltage fluctuation rate, load fluctuation rate, user voltage evaluation and user maximum voltage drop ratio; the economic dimension indicators include total electricity consumption and load rate.

3. The method for constructing a multi-dimensional feature portrait of a transformer area voltage based on ISSA clustering fusion according to claim 2 is characterized in that: The process of obtaining the user voltage evaluation includes the following steps: Step S201, constructing a standardized indicator matrix of the voltage deviation size, voltage fluctuation rate and power supply distance of users in the substation area, obtaining a probability matrix by calculating the indicator proportion, calculating the information entropy and information utility value of each indicator based on the probability matrix, and normalizing to obtain the entropy weight of the indicator; Step S202 , performing forward processing on the standardized indicator matrix to obtain a standardized extremely large indicator matrix, and obtaining a user voltage evaluation score based on the extremely large indicator matrix and the entropy weight.

4. The method for constructing a multi-dimensional feature portrait of a substation voltage based on ISSA clustering fusion according to claim 3 is characterized in that: The user voltage evaluation is calculated using the following formula: Among them, Y4 is the user voltage evaluation score, D i + is the distance between the ith index and the maximum value, D i - is the distance between the i-th index and the minimum value, Q Mj + is the maximum value in the standardized extremely large indicator matrix; Q Mj - is the minimum value in the standardized extremely large indicator matrix, ω j is the entropy weight of the jth indicator, and m is the number of indicators.

5. The method for constructing a multi-dimensional feature profile of a substation voltage based on ISSA clustering fusion according to claim 1 is characterized in that: The step S3 includes the following sub-steps: Step S301, using ISSA to initialize the center positions of candidate clusters and the number of clusters; Step S302: for each cluster, using K-means algorithm to assign voltage feature labels to the nearest cluster center; Step S303, calculating the fitness function value based on the cluster silhouette coefficient; Step S304, using the ISSA model with nonlinear operators and linear weights, iteratively obtains a new cluster center position, and repeats steps S302-S303 until the fitness function value satisfies the termination adjustment.

6. The method for constructing a multi-dimensional feature profile of a substation voltage based on ISSA clustering fusion according to claim 5, characterized in that: The ISSA model is: The dynamic update process of the follower's position is: Among them, t is the current iteration number, T max is the maximum number of iterations, For the leader position, F j is the target position, ub j lb j are the upper and lower limits of the exploration space, c1 is the change factor, c2 and c3 are random numbers between [0,1], α(t) is a nonlinear operator, and x i,j is the follower position; x i-1,j is the neighbor follower position and β is the linear weight.

7. The method for constructing a multi-dimensional feature profile of a transformer substation voltage based on ISSA clustering fusion according to claim 1, characterized in that: The following steps are also included: Step S4, taking the clustering result as a sample, inputting it into a pre-built BP neural network for iteration, and using the iterated BP neural network to realize the recognition of the voltage operation mode of the substation area, and the recognition of the voltage problem type existing in the substation area with abnormal power grid operation.

8. The method for constructing a multi-dimensional feature image of a substation voltage based on ISSA clustering fusion according to claim 1 is characterized in that: In the step S1, the voltage and load data of the sampling points are the voltage and load data of 96 sampling points per day. After obtaining the voltage and load data of the sampling points, the following steps are further included: The voltage and load data of the sampling points are subjected to missing value and abnormal value processing, data transformation, data integration and data reduction processing.

9. An electronic device, characterized in that: include: One or more processors and a memory, wherein the memory stores one or more programs, and the one or more programs include instructions for executing the method for constructing a multi-dimensional feature portrait of substation voltage based on fusion ISSA clustering as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that: It includes one or more programs executed by one or more processors of an electronic device, and the one or more programs include instructions for executing the method for constructing a multi-dimensional feature portrait of substation voltage based on fused ISSA clustering as described in any one of claims 1-8.

Images