Evaluation Index System Of Anti Stealing Electriity And Analysis Of Stealing

Stealing electricity has the bad social impact,not only broke the balance of supply and demand,but also seriously affect the economic benefits of power supply company,resulting in loss of state.At present,manual inspection and information collection are the mainly means of stealing electricity investigation by electricity electric power industry in our country.Low efficiency of manual inspection method and data distortion existed in information collecting,which is difficult to judge the truth of stealing electricity.So it is vital to find a more scientific and efficient investigation of stealing electricity.This paper mainly completed the following work:(1)The situation and method of anti stealing electricity in the domestic and foreign are analyzed.A large number of domestic and foreign related data used in other research is accumulated,which is convenient to later research work on load.(2)Load data of high line loss is chose.Then the pretreatment method of load data is put forward.Through the data pretreatment methods of data reduction and data elimination to process the original data from power supply company,the original monthly consumption data is covered into high quality data set which can be analyzed by data mining in the research.(3)Evaluation index system of stealing electricity established is to determine the suspicion users of stealing electricity.Data mining technology in the K-means clustering method is applied to analyze the user instance.Through distinguishing index feature and characteristic curve different from the normal users,evaluation index system of stealing electricity is established to determine the users suspected users.Then suspected users list will feed back to power supply company to examine on the spot.(4)Stealing electricity suspicion evaluation model is established to forecast the truth of stealing electricity users.Decision tree and artificial neural network method in data mining techniques are applied to establish the anti stealing evaluation model to forecast to judge whether the user stealing or not.This explains to some extent that data mining methods applied in stealing prediction is feasible and reliable.Finally,the application of the proposed method and the developed models are applied in real power supply load data acquisition.The stealing electricity users' lists from stealing electricity evaluation index system and evaluation model established in the paper are verified by field verification of power supply company.