Study on Tariff Risk Early Warning of Electric Power Users Based on PSO-SVM Algorithm

Tariff recovery risks of electricity power users are always major problems for electric power supply company, and data mining technology can realize early warning of tariff recovery risks by analyzing users' power consumption, payment and arrears data. Firstly, based on basic information, payment behavior, and default behavior of electric power users, this paper establishes a risk warning model for tariff recovery. Secondly, for the penalty factor C and RBF kernel parameter g of support vector machine (SVM) is directly related to the accuracy of recognition rate of overdue users, a particle swarm optimization (PSO)-support vector machine (PSO-SVM) algorithm is proposed to solve the risk warning model. The analysis results of test case show that the model and algorithm proposed in this paper have high availability and accuracy, and can effectively warn tariff recovery risks of electric power users.