Research On State Of Charge Estimation For Lythium-Ion Battery Used In Electric Vechile

State of charge of batteries which can't be measured directly is an important parameter for a battery, as it indicates the usable charge of a battery, it is an important parameter of the electric vehicle energy optimism algorithm. So SOC estimation method is one of the key techniques of the battery management system. This paper studied an SOC estimation method used to estimate the SOC of Lithium battery used in electric vehicles via sliding mode observer.First, a model based on Thevenin equivalent circuit model was established and the parameter was identified through the nonlinear least square method and battery discharge experiments including const-current discharge and pulse discharge. At last a nonlinear state space model of a battery was established.Then, the observability of the state space model was analyzed, and the result showed that the model was observable, so there exist a state observer. Therefore a sliding mode observer which can eliminate the jitter was designed, in order to analyze the stability of the observer the nonlinear feature of the battery model was analyzed, and the result showed that it had an bounded derivative. On this basis, the theorem of Lagrange's mean was used to design a stable observer. And the way to obtain the sliding mode observer parameter was designed, too. Subsequently, a simulation experiment was carried out using Matlab/Simulink. The SOC estimation method using Kalman filter was used, too, as a control experiment. The results showed that sliding mode observer method had better predicting performance comparing to the extended Kalman filter method when there exists the same modeling errors.Finally, the estimation algorithm was improved. Since the battery has a feature that the usable charge of a battery declines when the discharge current increases,which always happens in electric vehicle. Battery model was modified according to the Peukert equation in order to guarantee the estimation accuracy, the sliding mode observer was modified accordingly. Next, a simulation experiment was carried out using Matlab/Simulink. The SOC estimation method before improvement was used as a control experiment. The results showed that improved sliding mode observer method had better predicting performance when the usable charge changed with the discharge current.