Adaptive mutation particle swarm optimized BP neural network in state-of-charge estimation of Li-ion battery for electric vehicles

作者：Feng Jin^*; He Yong ling

来源：Bulgarian Chemical Communications, 2015, 47(3): 904-912.

• Adaptive mutation
• particle swarm optimization
• BP neural network
• state of charge

The state of charge (SOC) of Li-ion battery on electric vehicle (EV) is highly nonlinear. The randomly selected initial parameters of BP neural network can cause significant inaccuracy and long training time. In the study presented in this paper, an optimized BP neural network, with its initial parameters optimized by adaptive particle swarm optimization (PSO) algorithm, was used to estimate the Li-ion battery's state of charge (SOC). The performance on BP neural network estimation, as well as the optimized performance with adaptive mutation PSO was analyzed. A model for adaptive mutation PSO- BP neural network was established for battery SOC estimation. Experimental results show that: using BP neural network optimized by adaptive mutation PSO for SOC estimation of Li-ion battery of EV, can overcome the shortcomings of easily trapped to local optimum, long training time and so on. It also reduces the estimation deviation.

• 出版日期2015
• 单位北京大学; 桂林航天工业学院