Research On Distribution Network Loss And Reconstruction Based On Deep Learnin

With the gradual exhaustion of traditional energy sources,the development and research of new energy sources have attracted more and more attention of researchers.Network loss is usually a reference index to measure the operation level of power companies,and it is often to improve operating efficiency and save energy and reduce emissions.The goal can be achieved by reducing the network loss.Due to the continuous expansion of the scale of the distribution network,it is required that the accuracy of calculating the network loss must be high enough,so as to help the management and operation of the power grid company.Therefore,in this paper,considering the accuracy of network loss calculation,based on the distribution network model including distributed power generation and charging vehicles,the corresponding network reconfiguration and network loss calculation research are carried out.The main research results are as follows:(1)Due to the limited calculation method and unsatisfactory accuracy of the power grid line loss in the low-voltage station area,this paper firstly uses the scientific AHP to analyze the weight of the line loss index,which better replaces the traditional expert experience method.Then,for the shortage of initial data clustering processing,the improved K-Means unsupervised clustering algorithm based on the initial number of categories and cluster centers is applied.Finally,the regression and approximation ability of the linear model by Orthogonal Least Squares(OLS)is used to improve the performance of the traditional(Radical Basis Function,RBF)neural network,and the improved OLS-RBF model is applied In the power grid line loss calculation in a lowvoltage station area in a certain area,the feasibility and superiority of the improved model are shown through the analysis of the simulation results.The RBF model improved by OLS has a simpler structure,higher calculation accuracy and faster convergence speed.An effective algorithm for line loss value of power grid in Taiwan area.(2)Considering the access of a large number of new power sources and complex loads,the distribution network is no longer the original single power supply network,which makes the traditional distribution network reconstruction method no longer applicable.Therefore,considering the randomness and uncertainty of power flow,this paper adopts a three-step heuristic algorithm to solve this problem.The accuracy and speed are improved,and finally the simulation experiment proves that the accuracy and speed of the method are better than other algorithms.(3)Considering the frequent changes of the distribution network load and the randomness of the grid connection period of the distributed power generation,the realtime reconstruction of the distribution network is more important.In this paper,for the speed and efficiency of the real-time reconstruction of the distribution network,the A three-step heuristic algorithm is used to obtain the optimal reconstruction solution of the historical data,and then the convolutional neural network is used to establish the nonlinear mapping relationship between the optimal reconstruction solution and the corresponding real-time topology structure of the distribution network.According to the algorithm and the three-step heuristic algorithm mentioned in the previous chapter,the validity and correctness of the algorithm are obtained,and the scenarios under the abnormal conditions of DG and EV are also considered,which shows that the algorithm is fast and fast after a sudden failure.Restoring the power supply also has better adaptability.