Comparison of Four Intelligent Algorithms for Battery SOC Estimation in Electric Vehicle

It is of great importance to accurately estimate the battery State-Of-Charge (SOC) during electric vehicle driving process. Although much research work has been carried on in recent years, it still cannot be totally solved. Four intelligent algorithms with good nonlinear approximation ability are adopted to estimate the battery SOC, which are BP neural network (BPNN), Elman neural network (Elman), epsilon-SVR and nu-SVR. The simulation results show that, all the four algorithms can get good approximation to the actual value, and the average estimation error is less than 2%, while the estimation performance using nu-SVR algorithm is best.