Detection Method for Power Theft Based on SOM Neural Network and K-means Clustering Algorithm

With the diversification and complexity of power theft methods, the traditional detection methods are difficult to keep up with the changes of power theft methods. Therefore, the detection method of electricity theft based on data mining technology has become a hot research topic. SOM neural network and K-means clustering algorithm are commonly used clustering methods. SOM neural network can automatically determine the number of clusters, but cannot give accurate clustering information; meanwhile, K-means clustering algorithm has high accuracy but needs to give the initial value in advance. In this paper, on the basis of analyzing the principle of relevant algorithms and the characteristics of electricity data, a method for detecting electricity theft users based on two clustering algorithms is proposed, which can accurately identify electricity theft users through deep mining of abnormal data of electricity users. Theoretical analysis and experimental results show that the method can effectively improve the accuracy of identification of electricity theft, and has certain practicability.