A combined method of battery SOC estimation for electric vehicles

Abstract

Exact estimation of battery state of charge (SOC) is important for a monitoring system, which is the basis of a energy management system(EMS) in electric vehicles. This paper presents a combined method for estimating the battery SOC for electric vehicles. Diagonal recurrent neural network (DRNN) and Kalman filter (KF) were used to estimate battery SOC respectively. Then the two methods were combined to apply alternately. The combined method synthetized the advantages of the neural network and the Kalman filter, so the it can not only estimate SOC accurately, but also reduce computation amount.