Intelligent Identification Method for Intermittent Electricity Theft Behavior of Users Based on Multitask Learning

In order to improve the detection rate and AUC value of electricity theft behavior recognition results, and reduce the false detection rate, a user intermittent electricity theft behavior intelligent recognition method based on multi task learning is proposed. By collecting and analyzing user electricity consumption data, identify the characteristics of electricity theft behavior, obtain electricity theft feature data, and use factor analysis to dimensionally regulate the electricity theft feature data. Based on the results of data processing, train a multi task learning model to gain a more comprehensive understanding of users' electricity consumption and identify intermittent electricity theft behavior. The experimental results show that the detection rate and AUC value of the proposed method are higher, and the false detection rate is significantly reduced, indicating that the recognition effect of this method is good.