Early Warning Method of Tariff Recovery Risk Based on Long Short-Term Memory Neural Network

Abstract

The early warning of users' tariff recovery risk is a very difficult problem during the operation of electricity companies. This paper proposes a novel early warning method of tariff recovery risk based on the long short-term memory (LTSM) neural network. First, the concept of LTSM neural network is introduced. Then, the detailed procedure of early warning of tariff recovery risk based on LTSM neural network is proposed and described. At last, the proposed method is compared to other commonly used algorithms such as logistic regression, decision tree algorithm, and support vector machines on a set of practical data, and the results show the high efficiency of the proposed method.