battery with neural network method

 

 

CAI Xin, LI Bo, WANG Honghua, NIE Liang

 

(State Grid Zhejiang Electric Vehicle Company, Hangzhou, 310007, China)

 

 

Abstract: Aiming at estimating stateofcharge of electric vehicle batteries precisely, study on modeling stateofcharge(SOC) was investigated. After analyzing effect factors of SOC, a model of SOC on BP neural network was established. Electric vehicle simulation software ADVISOR was used to simulate an electric vehicle on typical driving cycles. The relationship between current, temperature and SOC was derived through simulation. After normalization of the training data and training the neural network, a SOC estimation model based on BP neural network was derived. The model was tested by testing data obtained by ADVISOR. The results indicate that the maximum error between the model estimation and actual values is 4%, which can satisfy the requirements of actual use of SOC.

 

Key words: electric vehicle; stateofcharge(SOC); neural network