Substation line loss prediction method and system considering electrical characteristic index optimization strategy

Abstract

The present invention discloses a method and system for predicting line loss in a transformer area by considering an electrical characteristic index optimization strategy. The method of the present invention comprises determining a prediction sample X i , the predicted sample X i Including a variety of specified electrical characteristics of the area to be predicted; calculate the prediction sample X i Any sample X in the historical sample data after feature optimization is completed in advance j The weighted Euclidean distance between the predicted sample X i The K samples with the smallest weighted Euclidean distance are selected; the line loss rate of the K samples found in the historical sample data after the feature optimization is completed and their relationship with the predicted sample X i The weighted Euclidean distance of the predicted sample X i The present invention solves the problem of difficult feature selection in traditional substation line loss prediction algorithms. It uses an index to determine the correlation between electrical characteristic indicators and line loss rates, implements feature optimization in high-dimensional features at the algorithm level, and uses an instance-based calculation model. This eliminates the need for training historical data, reduces model complexity, shortens calculation time, and simplifies calculations.

Description

Technical Field

[0001] The present invention relates to electrical measurement technology, and in particular to a method and system for predicting line loss in a transformer area by considering an electrical characteristic index optimization strategy. Background Art

[0002] Accurate prediction of the line loss rate in the substation area is an important basis for formulating loss reduction plans, providing a theoretical basis for optimizing the operation of the power grid, further strengthening the line loss management and transformation of the substation area, and ensuring that the power grid loss reduction is accurate and effective.

[0003] Traditional line loss theoretical calculation methods include the root mean square current method, the equivalent resistance method, and the maximum current method. These methods are simple to calculate, but they require simplified assumptions about the network, or data such as the network structure, distribution transformer, or line model to be obtained for calculation. In order to solve the difficulty in calculating theoretical line losses, machine learning algorithms have been introduced into the theoretical line loss prediction of substations. However, the existing line loss calculation models based on machine learning algorithms have problems such as dimensional disasters caused by redundant features and irrelevant features, weak generalization ability, and low model efficiency. That is, they ignore the feature selection link, resulting in high model complexity and long training time. Summary of the invention

[0004] Technical problem to be solved by the present invention: In view of the above-mentioned problems in the prior art, a method and system for predicting line loss in a substation considering the optimization strategy of electrical characteristic indicators are provided. The present invention solves the disadvantage of difficult feature selection in traditional substation line loss prediction algorithms, judges the correlation between electrical characteristic indicators and line loss rates through indexes, and realizes feature optimization in high-dimensional features at the algorithm level. By applying an instance-based calculation model, there is no need to train historical data, which reduces the complexity of the model, shortens the calculation time, and simplifies the calculation.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] A method for predicting line loss in a transformer area considering an electrical characteristic index optimization strategy includes:

[0007] 1) Determine the prediction sample X i , the predicted sample X i Includes multiple specified electrical characteristics of the area to be predicted;

[0008] 2) Calculate the predicted samples X respectively i Any sample X in the historical sample data after feature optimization is completed in advance j The weighted Euclidean distance between the predicted sample X i K samples with smaller weighted Euclidean distance;

[0009] 3) According to the line loss rate of K samples found in the historical sample data after feature optimization and its correlation with the predicted sample Xi The weighted Euclidean distance of the predicted sample X i Line loss rate.

[0010] Optionally, the calculation function expression of the weighted Euclidean distance in step 2) is:

[0011]

[0012] In the above formula, dist(i,j) is the predicted sample X i Compared with sample X in historical sample data j The weighted Euclidean distance between them, M′ is the number of samples of historical sample data after feature optimization, and N is the number of samples of historical sample data before feature optimization. For the prediction sample X i The mth electrical characteristic in For sample X j The mth electrical characteristic in .

[0013] Optionally, in step 3), the predicted sample X is calculated i The functional expression of line loss rate is:

[0014]

[0015] In the above formula, Represents the predicted sample X i Line loss rate, dist(i,j) is the predicted sample X i Compared with sample X in historical sample data j The weighted Euclidean distance between them, dist(i,k) is the predicted sample X i Compared with sample X in historical sample data k The weighted Euclidean distance between them, K is the weighted Euclidean distance between them and the predicted sample X i The weighted Euclidean distance of the smaller number of samples, y k is the sample X in the historical sample data k Line loss rate.

[0016] Optionally, before step 2), the step of generating historical sample data after feature optimization is included:

[0017] S1) collecting historical sample data, wherein the historical sample data includes N samples, each sample includes a feature vector and its corresponding line loss rate, wherein the feature vector is composed of M types of electrical features;

[0018] S2) All electrical characteristics and line loss rates of N samples are distributed on a two-dimensional plane, and the two-dimensional plane is divided into a grid with a rows and b columns. The electrical characteristics x of any category are calculated according to the grid. m The joint probability between the line loss rate y and the electrical characteristics xm , the marginal probability of line loss rate y, and calculate the electrical characteristics x of any category m The mutual information and correlation information index between y and line loss rate y;

[0019] S3) Filter various electrical features x whose associated information index is less than a preset associated information index threshold β for the original historical sample data m , thus obtaining the historical sample data after feature optimization.

[0020] Optionally, when the two-dimensional plane is divided into a grid of a rows and b columns in step S2), the values ​​of a and b satisfy:

[0021]

[0022] In the above formula, floor means rounding, and N is the number of samples in the historical sample data.

[0023] Optionally, in step S2), the electrical characteristics x of any category are calculated m The calculation function expression of the joint probability between and line loss rate y is:

[0024]

[0025] In the above formula, P(x m ,y) is the electrical characteristic x m The joint probability between and line loss rate y, max is the maximum value, N a,b is the number of samples falling within any grid (a, b), and N is the number of samples in the historical sample data; in step S2), the electrical characteristics x of any category are calculated m The calculation function expression of the marginal probability is:

[0026]

[0027] In the above formula, P(x m ) is the electrical characteristic x m The marginal probability of is the electrical characteristic x m The number of samples falling in a certain column of the grid; the calculation function expression of the edge probability of the line loss rate y in step S2) is:

[0028]

[0029] In the above formula, P(y) is the marginal probability of line loss rate y, N y is the number of samples of line loss rate y falling within a column of the grid.

[0030] Optionally, in step S2), the electrical characteristics x of any category are calculated mThe calculation function expression of the mutual information between and line loss rate y is:

[0031]

[0032] In the above formula, I[x m ; y] represents the electrical characteristic x m and the mutual information between the line loss rate y, where Y is the set of all line loss rates, P(x m ,y) is the electrical characteristic x m The joint probability between the line loss rate y, P(x m ) is the electrical characteristic x m , P(y) is the marginal probability of line loss rate y.

[0033] Optionally, in step S2), the electrical characteristic x m The calculation function expression of the correlation information index between and line loss rate y is:

[0034]

[0035] In the above formula, CorrInf[x m ; y] represents the electrical characteristic x m and the line loss rate y, I[x m ; y] represents the electrical characteristic x m and the line loss rate y, floor represents rounding, and N is the number of samples in the historical sample data.

[0036] In addition, the present invention also provides a substation line loss prediction system that considers the electrical characteristic index optimization strategy, including an interconnected microprocessor and a memory, and the microprocessor is programmed or configured to execute the steps of the aforementioned substation line loss prediction method that considers the electrical characteristic index optimization strategy.

[0037] In addition, the present invention also provides a computer-readable storage medium, which stores a computer program for being executed by a computer device to implement the aforementioned substation line loss prediction method considering the electrical characteristic index optimization strategy.

[0038] Compared with the prior art, the present invention mainly has the following advantages:

[0039] 1. The method of the present invention includes determining the prediction sample X i , the predicted sample X i Including a variety of specified electrical characteristics of the area to be predicted; calculate the prediction sample X i Any sample X in the historical sample data after feature optimization is completed in advance j The weighted Euclidean distance between the predicted sample X iThe K samples with the smallest weighted Euclidean distance are selected; the line loss rate of the K samples found in the historical sample data after the feature optimization is completed and their relationship with the predicted sample X i The weighted Euclidean distance of the predicted sample X i The present invention solves the problem of difficult feature selection in traditional substation line loss prediction algorithms. It uses an index to determine the correlation between electrical characteristic indicators and line loss rates, implements feature optimization in high-dimensional features at the algorithm level, and uses an instance-based calculation model. This eliminates the need for training historical data, reduces model complexity, shortens calculation time, and simplifies calculations.

[0040] 2. Calculate the predicted samples X separately i Any sample X in the historical sample data after feature optimization is completed in advance j The weighted Euclidean distance between the predicted sample X i The improved K nearest neighbor algorithm is simpler than the existing neural network model, easier to implement, and insensitive to outliers in the data set. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 Schematic diagram of the basic flow of the method of the embodiment of the present invention.

[0042] Figure 2 The figure is a schematic diagram of the process of generating historical sample data after feature optimization in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] like Figure 1 As shown, the method for predicting line loss in a transformer area considering the electrical characteristic index optimization strategy in this embodiment includes:

[0044] 1) Determine the prediction sample X i , the predicted sample X i Includes multiple specified electrical characteristics of the area to be predicted;

[0045] 2) Calculate the predicted samples X respectively i Any sample X in the historical sample data after feature optimization is completed in advance j The weighted Euclidean distance between the predicted sample X i K samples with smaller weighted Euclidean distance;

[0046] 3) According to the line loss rate of K samples found in the historical sample data after feature optimization and its correlation with the predicted sample X i The weighted Euclidean distance of the predicted sample X i Line loss rate.

[0047] In this embodiment, the calculation function expression of the weighted Euclidean distance in step 2) is:

[0048]

[0049] In the above formula, dist(i,j) is the predicted sample X i Compared with sample X in historical sample data j The weighted Euclidean distance between them, M′ is the number of samples of historical sample data after feature optimization, and N is the number of samples of historical sample data before feature optimization. For the prediction sample X i The mth electrical characteristic in For sample X j The model calculation of weighted Euclidean distance is simple and can be completed with simple MATLAB statements, which is easy to implement in engineering applications.

[0050] In this embodiment, the predicted sample X is calculated in step 3). i The functional expression of line loss rate is:

[0051]

[0052] In the above formula, Represents the predicted sample X i Line loss rate, dist(i,j) is the predicted sample X i Compared with sample X in historical sample data j The weighted Euclidean distance between them, dist(i,k) is the predicted sample X i Compared with sample X in historical sample data k The weighted Euclidean distance between them, K is the weighted Euclidean distance between them and the predicted sample X i The weighted Euclidean distance of the smaller number of samples, y k is the sample X in the historical sample data k Calculate the predicted sample X i The weighted summation method is used in the functional expression of the line loss rate. The weight coefficient of the summation is determined according to the distance between the sample points, so that the features with larger changes have smaller weight coefficients than the features with smaller changes, reducing random fluctuations and ensuring the accuracy of the prediction results. i The line loss rate model calculation is simple and can be completed using simple MATLAB statements, making it easy to implement in engineering applications.

[0053] like Figure 2 As shown, before step 2) of this embodiment, the step of generating historical sample data after feature optimization is completed is included:

[0054] S1) collecting historical sample data, the historical sample data includes N samples (the value of N can be specified as needed, for example, in this embodiment, the specific value of N is 2000), each sample includes a feature vector and its corresponding line loss rate, wherein the feature vector is composed of M types of electrical features;

[0055] In this embodiment, N samples of historical sample data are represented as:

[0056] {(X 1 ,y 1 ),(X 2 ,y 2 ),...,(X N ,y N )}

[0057] Among them, X 1 ,X 2 ,…,X N are the electrical characteristic vectors of N samples, y 1 ,y 2 ,…,y N are the line loss rates of N samples respectively, and the electrical characteristic vector X of any sample i is i It is expressed as:

[0058]

[0059] in, The electrical characteristic vectors X i In the embodiment, M=6, the M electrical characteristics include: the voltage drop at the head and the end, the power factor, the load rate, the proportion of the end electricity, the three-phase imbalance and the proportion of the online electricity. Since the above electrical characteristics are all well-known electrical characteristics, their specific physical meanings are not described in detail here. In this embodiment, some samples of the historical sample data are shown in Table 1.

[0060] Table 1: Some sample examples of historical sample data.

[0061]

[0062] S2) All electrical characteristics and line loss rates of N samples are distributed on a two-dimensional plane, and the two-dimensional plane is divided into a grid with a rows and b columns. The electrical characteristics x of any category are calculated according to the grid. m The joint probability between the line loss rate y and the electrical characteristics x m , the marginal probability of line loss rate y, and calculate the electrical characteristics x of any category m The mutual information and correlation information index between y and line loss rate y;

[0063] In this embodiment, all electrical characteristics and line loss rates of the 2000 samples are distributed on a two-dimensional plane, and the two-dimensional plane is divided into a grid with a rows and b columns. In this embodiment, the values ​​of a and b satisfy:

[0064]

[0065] In the above formula, floor means rounding, and N is the number of samples in the historical sample data. For example, in this embodiment, the values ​​of a and b are both 5, and the grid is divided into 5×5, that is, a grid of 5 rows and 5 columns. When traditional machine learning methods are used to calculate line loss rates, there is a problem of dimensionality disaster caused by not considering redundant features or irrelevant features. Even the most widely used neural network model currently has slow convergence speed, network structure and parameter selection difficulties, resulting in certain limitations in practical applications. The method of this embodiment cleverly uses the thresholds of a and b to limit grid division. By constructing a grid division to approximate the calculation of the joint probability and edge probability in the mutual information, the calculation of the correlation information index is greatly simplified.

[0066] In this embodiment, in step S2), the electrical characteristics x of any category are calculated. m The calculation function expression of the joint probability between and line loss rate y is:

[0067]

[0068] In the above formula, P(x m ,y) is the electrical characteristic x m The joint probability between and line loss rate y, max[·] is the maximum value, N a,b is the number of samples falling within any grid (a, b), that is, The number of occurrences; N is the number of samples in the historical sample data; that is, the frequency of occurrence of sample points in each grid (a, b) is calculated respectively, and the maximum value is taken as {x m ,y}. Calculate the joint probability of any class of electrical features x m The model calculation of the joint probability between x and line loss rate y is simple and can be completed with simple MATLAB statements, which is convenient for implementation in engineering applications. m The joint probability P(x m ,y) is 0.368.

[0069] In step S2), the electrical characteristics x of any category are calculated. m The calculation function expression of the marginal probability is:

[0070]

[0071] In the above formula, P(xm ) is the electrical characteristic x m The marginal probability of is the electrical characteristic x m The number of samples that fall into a column of the grid. The above formula represents the calculation of feature x m The edge probability is to calculate each column of feature points x in the grid (a, b) separately m The frequency of occurrence, take the maximum value as P(x m ) approximate value. Calculate the electrical characteristics x of any class m The edge probability model calculation is simple and can be completed with simple MATLAB statements, which is convenient for implementation in engineering applications. m The marginal probability P(x m ) is 0.423. The calculation function expression of the marginal probability of line loss rate y in step S2) is:

[0072]

[0073] In the above formula, P(y) is the marginal probability of line loss rate y, N y is the number of samples of line loss rate y falling within a column of the grid.

[0074] In this embodiment, in step S2), the electrical characteristics x of any category are calculated. m The calculation function expression of the mutual information between and line loss rate y is:

[0075]

[0076] In the above formula, I[x m ; y] represents the electrical characteristic x m and the mutual information between the line loss rate y, where Y is the set of all line loss rates, P(x m ,y) is the electrical characteristic x m The joint probability between the line loss rate y, P(x m ) is the electrical characteristic x m , P(y) is the marginal probability of line loss rate y. Calculate the electrical characteristics x of any category m The model calculation of the mutual information between the electrical characteristic x and the line loss rate y is simple and can be completed with simple MATLAB statements, which is convenient for implementation in engineering applications. m and the line loss rate y m ; y] has a value of 0.3816.

[0077] In this embodiment, the electrical characteristic x in step S2) m The calculation function expression of the correlation information index between and line loss rate y is:

[0078]

[0079] In the above formula, CorrInf[x m ; y] represents the electrical characteristic x m and the line loss rate y, I[x m ; y] represents the electrical characteristic x m The mutual information between x and line loss rate y, floor means rounding, and N is the number of samples in the historical sample data. The method of this embodiment cleverly uses the information correlation between feature variables to construct the correlation information formula of two variables, introduces the variable correlation information calculation method, sets the threshold to select features with higher correlation information, reduces the calculation time of the line loss rate calculation model, and ensures the accuracy of the calculation results. Moreover, the electrical feature x m The calculation model of the calculation function expression of the correlation information index between and the line loss rate y is simple, and the calculation can be completed using simple MATLAB statements, which is convenient for implementation in engineering applications. In this embodiment, the electrical characteristic x m and the line loss rate y. m ; y] has a value of 0.27.

[0080] S3) Filter various electrical features x whose associated information index is less than a preset associated information index threshold β for the original historical sample data m , thus obtaining the historical sample data after feature optimization.

[0081] By using the feature optimization method described in steps S1) to S3), features that have little impact on the results during the calculation process can be eliminated, thereby reducing the complexity of the model, reducing the time consumption of building the model, speeding up the calculation, and improving the generalization ability of the model.

[0082] In this embodiment, the correlation information index threshold β is set to 0.2, and the features below the threshold are regarded as irrelevant features and are removed. Finally, the correlation information index of each electrical feature calculated is shown in Table 2:

[0083] Table 2: Correlation information index table of various electrical characteristics.

[0084]

[0085] Through comparison in Table 2, it is found that the correlation information indexes of "terminal power proportion" and "online power proportion" are 0.014 and 0.023 respectively, which are lower than the correlation information index threshold β. Therefore, the two electrical features of "terminal power proportion" and "online power proportion" are eliminated to obtain the historical sample data after feature optimization.

[0086] On this basis, the steps 1) to 3) of the method for predicting line loss in a substation area considering the optimization strategy of electrical characteristic indexes in this embodiment can be used to implement the prediction of a given prediction sample X. i Prediction of line loss rate.

[0087] In this embodiment, the value of K in step 2) is 5, and the predicted samples X are calculated respectively. i Any sample X in the historical sample data after feature optimization is completed in advance j The weighted Euclidean distance between the predicted sample X i The five samples with the smallest weighted Euclidean distance are shown in Table 3.

[0088] Table 3: Sample X with prediction i The 5 sample tables with the smallest weighted Euclidean distance.

[0089] Pressure drop at the beginning and end Power Factor Load rate / % Three-phase unbalance / % Line loss rate / % 13.28 0.99 33.73 64.77 3.74 87.52 0.99 12.07 66.77 3.74 25.71 0.98 16.13 58.96 3.74 32.74 1.00 19.49 36.70 3.73 37.89 0.97 35.33 30.13 3.73

[0090] Based on the prediction sample X shown in Table 3 i The five samples with the smallest weighted Euclidean distance, the line loss rate and its difference with the predicted sample X i The weighted Euclidean distance of i The line loss rate is 3.73.

[0091] In summary, the method for predicting line loss in substations that considers the electrical characteristic index optimization strategy in this embodiment solves the shortcoming of difficult feature selection in traditional substation line loss prediction algorithms. The correlation between electrical characteristic indicators and line loss rates is judged by indexes, and feature optimization in high-dimensional features is realized at the algorithm level. An instance-based calculation model is applied, and there is no need to train historical data, which reduces the model complexity, shortens the calculation time, and is simple to calculate. Moreover, the feature optimization method proposed in the method of this embodiment can eliminate features that have little impact on the results during the calculation process, thereby reducing the model complexity, reducing the time consumption of establishing the model, speeding up the calculation, and improving the generalization ability of the model. Since the method of this embodiment uses weighted summation to calculate the line loss rate, the weight coefficient of the summation is determined according to the distance between the sample points, so that features with larger changes have smaller weight coefficients than features with smaller changes, reducing random fluctuations, thereby ensuring the accuracy of the prediction results.

[0092] In addition, this embodiment also provides a substation line loss prediction system that considers the electrical characteristic index optimization strategy, including an interconnected microprocessor and a memory, and the microprocessor is programmed or configured to execute the steps of the aforementioned substation line loss prediction method that considers the electrical characteristic index optimization strategy.

[0093] In addition, this embodiment also provides a computer-readable storage medium, which stores a computer program for being executed by a computer device to implement the aforementioned method for predicting line loss in a substation considering the electrical characteristic index optimization strategy.

[0094] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the instructions in the process Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0095] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for predicting line loss in a substation area considering an electrical characteristic index optimization strategy, characterized in that: include: 1) Determine the prediction sample X i , the predicted sample X i Includes multiple specified electrical characteristics of the area to be predicted; 2) Calculate the prediction samples separately X i Any sample in the historical sample data after feature optimization is completed in advance X j The weighted Euclidean distance between them is used to find the predicted sample X i The weighted Euclidean distance of K samples; 3) Based on the historical sample data after feature optimization K Line loss rate of samples and its difference with predicted samples X i The weighted Euclidean distance of the predicted sample is calculated X i Line loss rate; The calculation function expression of the weighted Euclidean distance in step 2) is: In the above formula, dist( i , j ) is the prediction sample X i Compared with the samples in historical sample data X j The weighted Euclidean distance between M′ The number of samples of historical sample data after feature optimization is completed. N The number of samples of historical sample data before feature optimization is completed. For the prediction sample X i The m Electrical characteristics, For sample X j The m Electrical characteristics; Step 3) Calculate the predicted sample X i The functional expression of line loss rate is: In the above formula, Represents the prediction sample X i Line loss rate, dist( i , j ) is the prediction sample X i Compared with the samples in historical sample data X j The weighted Euclidean distance between i , k ) is the prediction sample X i Compared with the samples in historical sample data X k The weighted Euclidean distance between K For the prediction sample X i The weighted Euclidean distance of the smaller number of samples, For samples in historical sample data X k Line loss rate.

2. The method for predicting line loss in a transformer area considering the optimization strategy of electrical characteristic indicators according to claim 1, characterized in that: Step 2) includes the steps of generating historical sample data after feature optimization: S1) Collect historical sample data, the historical sample data includes N samples, each sample includes a feature vector and its corresponding line loss rate, where the feature vector is M Class electrical characteristics composition; S2) N All electrical characteristics and line loss rates of samples are distributed on a two-dimensional plane, which is divided into a OK b A grid of columns from which electrical characteristics of any type can be calculated x m and line loss rate y The joint probability and electrical characteristics between x m , Line loss rate y , calculating the marginal probability of any class of electrical characteristics x m and line loss rate y Mutual information and correlation information index between them; S3) Filter the original historical sample data whose associated information index is less than the preset associated information index threshold β Various electrical characteristics x m , thus obtaining the historical sample data after feature optimization.

3. The method for predicting line loss in a transformer area considering the optimization strategy of electrical characteristic indicators according to claim 2, characterized in that: In step S2), the two-dimensional plane is divided into a OK b When the column grid is a and b The value of satisfies: In the above formula, floor means rounding. N is the number of samples in the historical sample data.

4. The method for predicting line loss in a transformer area considering the optimization strategy of electrical characteristic indicators according to claim 2, characterized in that: Step S2) calculates the electrical characteristics of any category x m and line loss rate y The calculation function expression of the joint probability between is: In the above formula, P ( x m , y ) is the electrical characteristic x m and line loss rate y The joint probability between ] is the maximum value, N a,b For any grid ( a , b ), N is the number of samples in the historical sample data; in step S2), the electrical characteristics of any category are calculated x m The calculation function expression of the marginal probability is: In the above formula, P ( x m ) is the electrical characteristic x m The marginal probability of Electrical characteristics x m The number of samples falling in a column of the grid; Line loss rate in step S2) y The calculation function expression of the marginal probability is: In the above formula, P ( y ) is the line loss rate y The marginal probability of N y Line loss rate y The number of samples that fall within a column of the grid.

5. The method for predicting line loss in a transformer area considering the optimization strategy of electrical characteristic indicators according to claim 4 is characterized in that: Step S2) calculates the electrical characteristics of any category x m and line loss rate y The calculation function expression of the mutual information between is: In the above formula, I [ x m ; y ] indicates electrical characteristics x m and line loss rate y The mutual information between Y is the set of all line loss rates, P ( x m , y ) is the electrical characteristic x m and line loss rate y The joint probability between P ( x m ) is the electrical characteristic x m The marginal probability of P ( y ) is the line loss rate y The marginal probability of .

6. The method for predicting line loss in a transformer area considering the optimization strategy of electrical characteristic indicators according to claim 5, characterized in that: Step S2) Electrical characteristics x m and line loss rate y The calculation function expression of the correlation information index between is: In the above formula, CorrInf[ x m ; y ] indicates electrical characteristics x m and line loss rate y The correlation information index between I [ x m ; y ] indicates electrical characteristics x m and line loss rate y The mutual information between them, floor means rounding, N is the number of samples in the historical sample data.

7. A system for predicting line loss in a substation area considering an electrical characteristic index optimization strategy, comprising a microprocessor and a memory connected to each other, characterized in that: The microprocessor is programmed or configured to execute the steps of the method for predicting line loss in a substation considering the electrical characteristic index optimization strategy as described in any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program for being executed by a computer device to implement the method for predicting line loss in a substation area considering the electrical characteristic index optimization strategy as described in any one of claims 1 to 6.

Images