Abstract:The user's abnormal power consumption behavior will lead to abnormal growth of power load,causing huge losses to the power company,and also causing failure of the transmission line,thereby endangering public safety (such as fire,electric shock,etc.),and the conventional manual inspection is inefficient and has low precision.Aiming at the above problems,this paper first analyzes the behavior characteristics of abnormal power users,then selects the best feature indicators for training models,and finally,an improved nonlinear abnormal electricity behavior detection algorithm based on improved nonlinear weight adjustment particle swarm optimization algorithm BP neural network (NWPSO-BP) is established.The experimental results show that the proposed model has higher error convergence speed is compared with the existing abnormal electricity behavior detection.