Power line loss prediction method based on genetic algorithm
Through the power line line loss prediction method based on genetic algorithm, key electrical characteristic parameters are screened and the line loss rate prediction model is constructed, which solves the problem of inaccurate evaluation of power line line loss rate in the existing technology, and achieves higher prediction accuracy and power network optimization effect.
Patent Information
- Application Number
- CN202411646261.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-18
- Publication Date
- 2025-04-25
AI Technical Summary
It is difficult for the prior art to accurately evaluate the line loss rate of power lines, which affects the accuracy of power network planning, design and operation management.
The power line line loss prediction method based on genetic algorithm is adopted. By collecting electrical characteristic parameters, the slope gray correlation analysis method is used to screen key parameters affecting the line loss rate, and a power line line loss rate prediction model is constructed.
It improves the accuracy of the prediction of the line loss rate of the power line, reduces redundant information, and can more accurately predict the loss of the power line, thereby formulating an effective power network optimization strategy, reducing line loss, and improving the operating efficiency and reliability of the power network.
Smart Images
Figure CN119886399A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of electric power engineering, and in particular to a method for predicting power line line loss based on a genetic algorithm. Background Art
[0002] With the continuous development of social economy, electricity has become one of the important resources for daily life and industrial production. As an important transmission carrier of the power network, the safety and efficiency of power lines are important factors affecting power safety. Under the influence of natural factors and their own materials, power lines have certain line losses. Therefore, it is extremely important to effectively evaluate the line loss of power lines. Industry experts use line loss rate as an important indicator to characterize the planning, design, and operation management level of power networks, which is also extremely important for daily power line maintenance. The so-called line loss rate of power lines refers to the ratio of the power loss of power lines to the actual power supply. In order to effectively evaluate, some scholars have constructed models by analyzing parameters such as transformer capacity, wiring method, and load factor; using BP neural network models for comprehensive prediction to improve the prediction accuracy of line loss rate, but due to the different scales of research, the influencing factors are difficult to meet the actual power line loss evaluation. Summary of the invention
[0003] The purpose of the present invention is to overcome the shortcomings of the prior art that the power line loss rate is difficult to accurately evaluate, which affects the actual evaluation accuracy and has an adverse impact on power network planning and design, operation and management, and to provide a power line loss prediction method based on genetic algorithm.
[0004] The purpose of the present invention is achieved through the following technical solutions: The power line loss prediction method based on genetic algorithm includes the following steps: Step 1: Collect electrical characteristic parameters and line loss rate in the power line, and perform data cleaning on the electrical characteristic parameters; Step 2: Analyze the cleaned electrical characteristic parameters using the slope grey correlation analysis method, determine the correlation between each electrical characteristic parameter and the line loss rate according to the geometric shape similarity of the sequence curve, and select several electrical characteristic parameters that have the greatest impact on the line loss rate; Step 3, using several electrical characteristic parameters that have the greatest impact on the line loss rate as input and the power line line loss rate as output to construct a power line line loss rate prediction model based on a genetic algorithm; Step 4: predict the power line loss rate through the power line loss rate prediction model, and formulate corresponding power network optimization strategy according to the prediction result.
[0005] The line loss of power lines mainly includes the following categories: loss of power transmission lines, dielectric loss of capacitors and related cables, loss of iron cores or inner coils, etc. In the past, the prediction of line loss of distribution lines often relied on expert experience to select certain parameters, which may have certain deviations from the actual line loss of distribution networks. Therefore, this scheme uses the slope grey correlation analysis method to select line loss index parameters. The slope grey correlation analysis method mainly judges the sequence association based on the geometric similarity of the sequence curves. If the curves are closer, the corresponding sequence association degree will be greater, and vice versa, the association degree will be smaller.
[0006] In this scheme, the power line loss rate prediction model is based on genetic algorithms. With the development of power networks, the amount of data generated by power systems has increased dramatically. Genetic algorithms perform well in processing large-scale data sets and can quickly converge to a better solution, which is suitable for real-time or quasi-real-time line loss prediction needs.
[0007] Preferably, the electrical characteristic parameters in the power line include monthly active power supply, monthly reactive power supply, line length, monthly reactive power supply of public transformers, line cross-sectional area, proportion of high-loss transformers, capacity of reactive compensation device and line material.
[0008] Preferably, the step 2 is specifically as follows: The data set defining the electrical characteristic parameters is Xi (i=1, 2, 3, ..., n), and the data set defining the line loss rate is Y; According to the slope grey correlation analysis method, the geometric shape similarity of data set Y and data set Xi at each point is compared and the correlation coefficient is calculated; For each electrical characteristic parameter in the data set Xi, the correlation coefficient is weighted averaged to obtain the correlation value between the electrical characteristic parameter and the line loss rate; The electrical characteristic parameters are sorted according to the correlation values, and several characteristic parameters with a high correlation with the line loss rate are screened out.
[0009] Preferably, the step 3 is specifically as follows: S1: Several electrical characteristic parameters that have the greatest impact on line loss rate are divided into corresponding test and training data sets; S2: Use the optimized genetic algorithm to select the threshold range of the neural network learning rate and the number of hidden layers, and perform initialization settings; S3: Perform initialization data set analysis, set the corresponding number of iterations, calculate fitness, and arrange them in order; S4: Calculate and analyze the fitness value according to the result of step S3; S5: According to the set threshold range, set the corresponding safe area and dangerous area, and analyze the best and worst position values according to the data set’s own data; S6: Calculate the number of iterations. If the number reaches the threshold, proceed to step S7. If not, continue iterating. S7: After the number of iterations is completed, the global optimal data individual analysis is performed to obtain the optimal parameters optimized by the genetic algorithm.
[0010] As a preferred method, the power line loss prediction method based on genetic algorithm also evaluates and optimizes the power line loss rate prediction model, specifically: The root mean square error or absolute percentage error is used to evaluate the model prediction results, and the neural network parameters are adjusted according to the evaluation results to retrain the power line loss rate prediction model.
[0011] Preferably, in step 4, the power network optimization strategy specifically includes load management optimization, voltage management optimization, grid structure optimization and renewable energy integration.
[0012] The beneficial effects of the present invention are as follows: by collecting electrical characteristic parameters and line loss rates in power lines, and performing data cleaning and slope grey correlation analysis on these parameters, the method can screen out several electrical characteristic parameters that have the greatest impact on the line loss rate. This helps to reduce redundant information and improve the accuracy and efficiency of the prediction model. The use of genetic algorithms to construct a power line line loss rate prediction model can more accurately predict the loss of power lines. Based on these prediction results, corresponding power network optimization strategies can be formulated, such as load management optimization, voltage management optimization, grid structure optimization and renewable energy integration, thereby reducing line losses and improving the operating efficiency and reliability of the power network. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] Figure 1 is a flow chart of the present invention; Figure 2 It is a comparison chart between the slope grey correlation analysis method and the traditional correlation method; Figure 3 This is a comparison chart of line loss power prediction results; Figure 4 This is a comparison chart of absolute percentage errors under different lines. DETAILED DESCRIPTION
[0014] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more comprehensive and complete and fully convey the concept of the example embodiments to those skilled in the art.
[0015] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present application. However, those skilled in the art will appreciate that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, known methods, devices, realizations or operations are not shown or described in detail to avoid blurring the various aspects of the application.
[0016] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.
[0017] Example: Power line loss prediction method based on genetic algorithm, such as Figure 1 As shown, the following steps are included: Step 1: Collect electrical characteristic parameters and line loss rate in the power line, and perform data cleaning on the electrical characteristic parameters; Step 2: Analyze the cleaned electrical characteristic parameters using the slope grey correlation analysis method, determine the correlation between each electrical characteristic parameter and the line loss rate according to the geometric shape similarity of the sequence curve, and select several electrical characteristic parameters that have the greatest impact on the line loss rate; Step 3, using several electrical characteristic parameters that have the greatest impact on the line loss rate as input and the power line line loss rate as output to construct a power line line loss rate prediction model based on a genetic algorithm; Step 4: predict the power line loss rate through the power line loss rate prediction model, and formulate corresponding power network optimization strategy according to the prediction result.
[0018] The line loss of power lines mainly includes the following categories: loss of power transmission lines, dielectric loss of capacitors and related cables, loss of iron cores or inner coils, etc. In the past, the prediction of line loss of distribution lines often relied on expert experience to select certain parameters, which may have certain deviations from the actual line loss of distribution networks. Therefore, this scheme uses the slope grey correlation analysis method to select line loss index parameters. The slope grey correlation analysis method mainly judges the sequence association based on the geometric similarity of the sequence curves. If the curves are closer, the corresponding sequence association degree will be greater, and vice versa, the association degree will be smaller.
[0019] In this scheme, the power line loss rate prediction model is based on genetic algorithms. With the development of power networks, the amount of data generated by power systems has increased dramatically. Genetic algorithms perform well in processing large-scale data sets and can quickly converge to a better solution, which is suitable for real-time or quasi-real-time line loss prediction needs.
[0020] The electrical characteristic parameters in the power line include monthly active power supply, monthly reactive power supply, line length, monthly reactive power supply of public transformers, line cross-sectional area, proportion of high-loss transformers, capacity of reactive compensation device and line material.
[0021] The results calculated by the traditional correlation method are: monthly active power supply, total capacity of public transformers, monthly reactive power supply, line length, line cross-sectional area, reactive compensation device capacity, line material, etc. Therefore, by comparing the slope grey correlation analysis method with the traditional correlation method, the order of the main influencing factors is different. According to different characteristic parameters and correlation analysis results, the average prediction error analysis is carried out, and the corresponding results are obtained as follows Figure 2 As shown. Comprehensively compared, the line loss prediction analysis obtained by using the slope grey correlation analysis has a lower error and is closer to the actual usage. Several characteristic parameters with the greatest impact on the line loss rate are selected for research and analysis.
[0022] The step 2 is specifically as follows: The data set defining the electrical characteristic parameters is Xi (i=1, 2, 3, ..., n), and the data set defining the line loss rate is Y; According to the slope grey correlation analysis method, the geometric shape similarity of data set Y and data set Xi at each point is compared and the correlation coefficient is calculated; For each electrical characteristic parameter in the data set Xi, the correlation coefficient is weighted averaged to obtain the correlation value between the electrical characteristic parameter and the line loss rate; The electrical characteristic parameters are sorted according to the correlation values, and several characteristic parameters with a high correlation with the line loss rate are screened out.
[0023] The step 3 is specifically as follows: S1: Several electrical characteristic parameters that have the greatest impact on line loss rate are divided into corresponding test and training data sets; S2: Use the optimized genetic algorithm to select the threshold range of the neural network learning rate and the number of hidden layers, and perform initialization settings; S3: Perform initialization data set analysis, set the corresponding number of iterations, calculate fitness, and arrange them in order; S4: Calculate and analyze the fitness value according to the result of step S3; S5: According to the set threshold range, set the corresponding safe area and dangerous area, and analyze the best and worst position values according to the data set’s own data; S6: Calculate the number of iterations. If the number reaches the threshold, proceed to step S7. If not, continue iterating. S7: After the number of iterations is completed, the global optimal data individual analysis is performed to obtain the optimal parameters optimized by the genetic algorithm.
[0024] The power line loss prediction method based on genetic algorithm also evaluates and optimizes the power line loss rate prediction model, specifically: The root mean square error or absolute percentage error is used to evaluate the model prediction results, and the neural network parameters are adjusted according to the evaluation results to retrain the power line loss rate prediction model.
[0025] In step 4, the power network optimization strategy specifically includes load management optimization, voltage management optimization, grid structure optimization and renewable energy integration.
[0026] Taking the power distribution network in a certain area as an example, the evaluation method of this embodiment is compared with the traditional line loss rate evaluation method. On the basis of utilizing the slope grey correlation method, the corresponding characteristic parameters are extracted as the input of the genetic algorithm optimization evaluation model. At the same time, the root mean square error, absolute percentage error, and mean absolute percentage error are introduced as evaluation criteria for the two methods.
[0027] There are 600 power line data in the distribution power network, so the corresponding data sampling interval is set to 15 minutes and the scale is 1 meter. The 600 line data are evenly divided into 5 sub-sample sets, and cross-validation is performed, taking turns to use each group of sub-sample data sets as test data sets, and the remaining 4 groups as test sample data sets, so the whole cycle is performed 5 times. In order to better illustrate the superiority and accuracy of the genetic algorithm optimization evaluation model, it is compared with the traditional evaluation model.
[0028] The calculation results of the absolute percentage error and root mean square error of the two methods are shown in Table 1 and Table 2. From the final calculation results, it can be seen that the genetic algorithm optimization evaluation model has higher prediction accuracy than the traditional evaluation model. The absolute percentage error of the 5 test data sets is only about 4%. Among the data sets with the largest error, the maximum absolute percentage error is only 4.22%. The larger the root mean square value, the greater the gap between the predicted evaluation result and the actual result. From the calculation results, compared with the traditional evaluation model, the root mean square value of the evaluation model optimized by the genetic algorithm is smaller, indicating that it has a better distribution network line loss rate evaluation effect.
[0029] Table 1 Absolute percentage errors of different models for 5 datasets Table 2 Root mean square error of different models for 5 data sets In order to reflect the actual performance of the genetic algorithm optimization evaluation model in the prediction of power network rate, 80% of the 600 lines are used as training samples, and the remaining 20% are used as test sample data sets. The line loss rate evaluation results of some lines selected as test samples are shown in Table 3: Table 3 Error comparison of two methods for some lines From the actual results, it can be seen that the power loss curve of the evaluation model optimized by genetic algorithm is closest to the actual power loss curve. Figure 3 shown.
[0030] Judging from the results of the power network line loss rate, the curve predicted by the genetic algorithm optimization model has the highest fit with the actual line loss rate prediction curve. This is mainly because the genetic algorithm optimization adds 4 door units on the basis of the original genetic algorithm, which further increases the relative interaction between the input and output of the internal model after genetic algorithm optimization, further improves the approximation ability of the model, and is more conducive to the prediction and evaluation of the line loss rate of the distribution power network.
[0031] By comparing the line loss quantity and the line loss rate, it can be seen that the 57th line has a low power network line loss quantity and a high relative line loss rate, while the 50th line has a high power network line loss quantity and a low relative line loss rate. Therefore, after comparative analysis, it can be seen that from the perspective of specific electrical characteristic parameters, this is because the relative capacity of the public transformer of the 57th line is relatively high, while the corresponding line length is relatively short. This also represents that the electrical characteristic parameters selected in Section 2 of this study can not only perform a good fitting analysis of line loss, but also reflect the actual operation of the power network distribution network, which further proves the scientificity and effectiveness of the slope grey correlation method in selecting characteristic parameters.
[0032] In order to more directly compare the error between the line loss prediction value and the actual measured value, the absolute percentage error of the two methods is calculated respectively. The specific results are as follows: Figure 4 shown.
[0033] from Figure 4 The results show that the absolute percentage error of the genetic algorithm optimization method proposed in this study is much lower than that of the traditional evaluation method, and the absolute percentage error is controlled to be no more than 8%. The largest absolute percentage error occurs on 102 lines, with a value of 7.84%.
[0034] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the embodiments disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application.
[0035] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.
Claims
1. A power line loss prediction method based on genetic algorithm, characterized in that: The following steps are involved: Step 1: Collect electrical characteristic parameters and line loss rate in the power line, and perform data cleaning on the electrical characteristic parameters; Step 2: Analyze the cleaned electrical characteristic parameters using the slope grey correlation analysis method, determine the correlation between each electrical characteristic parameter and the line loss rate according to the geometric shape similarity of the sequence curve, and select several electrical characteristic parameters that have the greatest impact on the line loss rate; Step 3, using several electrical characteristic parameters that have the greatest impact on the line loss rate as input and the power line line loss rate as output to construct a power line line loss rate prediction model based on a genetic algorithm; Step 4: predict the power line loss rate through the power line loss rate prediction model, and formulate corresponding power network optimization strategy according to the prediction result.
2. The method for predicting power line loss based on genetic algorithm according to claim 1, characterized in that: The electrical characteristic parameters in the power line include monthly active power supply, monthly reactive power supply, line length, monthly reactive power supply of public transformers, line cross-sectional area, proportion of high-loss transformers, capacity of reactive compensation device and line material.
3. The method for predicting power line loss based on genetic algorithm according to claim 1, characterized in that: The step 2 is specifically as follows: The data set defining the electrical characteristic parameters is Xi (i=1, 2, 3, ..., n), and the data set defining the line loss rate is Y; According to the slope grey correlation analysis method, the geometric shape similarity of data set Y and data set Xi at each point is compared and the correlation coefficient is calculated; For each electrical characteristic parameter in the data set Xi, the correlation coefficient is weighted averaged to obtain the correlation value between the electrical characteristic parameter and the line loss rate; The electrical characteristic parameters are sorted according to the correlation values, and several characteristic parameters with a high correlation with the line loss rate are screened out.
4. The method for predicting power line loss based on genetic algorithm according to claim 3 is characterized in that: The data of the data set of electrical characteristic parameters are normalized.
5. The method for predicting power line loss based on genetic algorithm according to claim 1, characterized in that: The step 3 is specifically as follows: S1: Several electrical characteristic parameters that have the greatest impact on line loss rate are divided into corresponding test and training data sets; S2: Use the optimized genetic algorithm to select the threshold range of the neural network learning rate and the number of hidden layers, and perform initialization settings; S3: Perform initialization data set analysis, set the corresponding number of iterations, calculate fitness, and arrange them in order; S4: Calculate and analyze the fitness value according to the result of step S3; S5: According to the set threshold range, set the corresponding safe area and dangerous area, and analyze the best and worst position values according to the data set’s own data; S6: Calculate the number of iterations. If the number reaches the threshold, proceed to step S7. If not, continue iterating. S7: After the number of iterations is completed, the global optimal data individual analysis is performed to obtain the optimal parameters optimized by the genetic algorithm.
6. The method for predicting power line loss based on genetic algorithm according to claim 5, characterized in that: The power line loss rate prediction model is also evaluated and optimized, specifically: The root mean square error or absolute percentage error is used to evaluate the model prediction results, and the neural network parameters are adjusted according to the evaluation results to retrain the power line loss rate prediction model.
7. The method for predicting power line loss based on genetic algorithm according to claim 1, characterized in that: In step 4, the power network optimization strategy specifically includes load management optimization, voltage management optimization, grid structure optimization and renewable energy integration.