Research On Line Loss Analysis Based On Big Data Theory

Line loss is closely related to the operation and management of the power grid in a distric t.Prec is e predic tion of the line loss has important guidance to the plan and management of the power grid.Through accurate recognition of abnormal line loss and according measures taken c an reduc e line loss and thus avoid unnec essary economic loss.With the promotion of smart meters and the improvement of user's information collection system,the era of big data in the power grid has come and thus power data resourc es begin to grow rapidly.Although traditional line loss prediction method has high accuracy,the prediction accurac y is greatly affected in the cas e of large data miss ing and big data set.The exis ting methods of recognizing electric ity larceny are mainly based on manpower,whic h has limitation on timeliness and accuracy.Therefore,how to apply the big data theory to the line loss prediction and abnorm detection to solve the problems existing in the traditional method has important research s ignific anc e.According to the background mentioned above,this paper focuses on the power line loss analys is method based on the big data theory,and mainly inc ludes the following work which has completed:(1)Research on the line loss calc ulation,line loss prediction,electric ity larceny recognition and the research status of the applic ation of big data theory to line loss analysis.(2)By comparing the calc ulation res ult of the existing theoretic al line loss calculation methods(such as root mean square current method,equivalent current method,line loss calc ulation method based on power flow calculation and optimal power flow method based on genetic algorithm),the different applicable s ituations of theoretic al line loss calculation method are proposed.(3)In the s ituation of complete data set,the trend of line loss is shown based on the improved EEMD and Elman neural network,the model of line loss prediction is established.Then the Elman neural network is trained.Compared with the s ingle predic tion model,the accuracy is significantly improved.(4)In the situation of data miss ing,the line loss prediction model is establis hed by using the compression sens ing method in the big data theory.The example shows that the prediction model proposed in this paper can solve the problem that data recovery with the traditional method can not reflect the trend of missing data and the prediction accuracy is not high.(5)Based on the high-dimens ional random matrix theory in big data theory,the high-dimens ional random matrix of both time and space dimens ion of operation data in the power grid is establis hed.Based on the matrix eigenvalue distribution and the variation of eigenvalue spectral dens ity with time,the recognition of electric larc eny occurrence,the measure to recognize the time period of electric larceny,the prec is e location of electric larceny and the type of elec tric larceny are proposed.Finally,examples are given to verify the accuracy and feasibility of the method.With the coming of the information age and the development of s mart grid,the amount of data available in the power grid will surge and the data dimens ion will continue to increase.Big data analys is methods have the advantages such as large extens ibility of data,strong visualization,low requirement of data integrity and high accurac y will become the main researc h field of future line loss analys is and lay the foundation for lean management of future power system.