Research On Clustering And Calculation Method Of Line Loss In Transformer District Based On Data Mining Technology

With the continuous development of power grid,our country pays more and more attention to the research of loss reduction and energy saving.The line loss rate of the distribution network is an important economic indicator to measure the production and operation of the power supply companies.The line loss rate can not only reflect whether the distribution network structure is reasonable,but also reflect the operation and management level of power supply enterprises.In the current distribution network system,there is a huge space for loss reduction in the low-voltage station area.However,due to the complex structure of the grid structure and the difficulty of obtaining original data,the traditional line loss calculation method is not accurate.In addition,the level of management personnel in the low voltage station area is generally not high,and there are phenomena such as estimating and missing copying,so it is difficult to implement unified management of line losses in the low voltage station area.Therefore,it is of great significance to implement lean management according to the characteristics of the station area,and to develop a method that can reliably and quickly calculate the theoretical line loss rate of the station area.In this paper,an improved K-Means clustering algorithm and GA-optimized LMBP neural network model are proposed to calculate the line loss rate of low-voltage stations,and an example is analyzed.Firstly,the original data are preprocessed to improve the data quality,and the characteristic indexes that have a greater impact on the line loss rate of the platform area are screened out by grey correlation analysis,Through the collinearity test,the characteristic indexes are combined to construct the characteristic system of the low-voltage stations.Secondly,considering the characteristics of the diverse user composition,this paper proposes an improved K-Means algorithm DWK to classify the transformer district,which effectively solves the problem of inaccurate calculation of the line loss rate caused by the large difference in the sample characteristics of the low-voltage stations.Then,the model based on GA-LMBP neural network was bulilt to learn and fit the complex nonlinear relationship between the line loss rate and the electrical characteristic parameters.So far,the theoretical line loss rate calculation models of various low-voltage stations were obtained;Finally,the sample data of 711 low-voltage stations of Wu an power supply company are selected to simulate the proposed model.The experimental results show that in terms of clustering,the improved DWK clustering algorithm has better clustering effect and applicability than traditional K-Means and other clustering algorithms.Compared with the BP neural network optimized by a single LM algorithm and the standard BP neural network,the GA-optimized LMBP neural network model has higher computational accuracy and convergence speed,which verifies the accuracy and feasibility of the method proposed in this paper.The research method in this paper makes up for the defect of insufficient application of power characteristic data in the traditional line loss management.By quickly and accurately calculating the line loss rate of each type of station area,it provides a reference for the line loss management of low voltage station area,and has practical engineering significance.