Electric power stealing behavior analysis and identification method based on electricity feature extraction

Anti-stealing electricity is the focus of power management in power supply enterprises. In view of the problems of increasing power theft and insufficient detection manpower in recent years, this paper proposes an analysis and identification method of power theft behavior based on power consumption feature extraction to screen suspected users of power theft. In this method, filter algorithm and rule threshold are used to extract and collect the features of big data samples, and a diagnostic model of users’ electric theft behavior is built based on logistic regression algorithm to judge suspected users of electric theft. The iterative optimization of the judgment model can be realized through the closed-loop working mechanism of checking and feedback. Finally, the feasibility of this method is verified by analyzing the data provided by the electricity information acquisition system of the power company The identification result of this method is high precision, which is suitable for the actual power consumption management of electric power enterprises. It is an effective means to prevent power theft in electric power enterprises, promote the sustainable development of enterprises and improve the power supply level.