Research And Implementation Of User Abnormal Power Detection System Based On Data Mining Technology

With the rapid spread of the power grid,residents enjoy the convenience of electricity consumption,while the power transmission enterprise’s line loss rate remains high,leading to an increase in the company’s power supply costs.The abnormal power consumption behavior of users is often the main reason for the high loss rate of power supply enterprises,which seriously affects the normal operation order of enterprises.Today,with the increasing degree of informationization of power grids,it is a hot research field to correctly use the massive data provided by the information grid to mine valuable information from a large amount of power information to solve the problem of abnormal power usage.This paper proposes an abnormal power detection model based on time dimension and principal component analysis.The model includes data preprocessing,time dimension based feature construction,principal component analysis,xgboost,GBDT,and random forest model.Then design and implement the abnormal electricity identification system based on the algorithm model.Compared with the traditional machine learning method,this project focuses on characterizing the user’s electricity usage behavior by constructing characteristics of different time dimensions from the original power load data.Then,the principal component analysis method is used to reduce the dimension of the high-dimensional features,so that the features used for training have different representations and a fusion model replaces traditional single model.In the verification phase of the model algorithm,this paper tests by using the actual power load data provided by the China Power Grid.Comparing the evaluation results of three single models with the fusion model used in this paper.It is found that the ability of the fusion model to detect abnonnal power users is significantly higher than that of the three single models.Moreover,the accuracy of the detection of abnormal power users or normal users is relatively high,and the method of model fusion is used to effectively improve the detection capability of abnormal power users.At the same time,this paper also designed and implemented the user abnormal power detection system.The system realizes the functions of data preprocessing,feature construction,model training,model detection result display,user management,etc.It can better identify abnormal power users and display the characteristics of power usage.