电动汽车用动力电池组SOC的神经网络估计

Neural network estimation of battery pack SOC for electric vehicles

针对电动汽车用动力电池组的SOC受充放电率、放电历程和温度等因素的影响,传统方法很难建立准确的数学模型,对电池组SOC进行研究,在对动力电池组进行不同工况充放电试验的基础上,建立了电池组的神经网络仿真模型。并分别采用电流输入,电压和电压梯度输入进行了仿真,实现了对电池组SOC的估计。与实验结果对比,仿真结果与实验基本吻合,验证了该方法的正确性。

It is very difficult to establish an accurate mathematical model to estimate the state of charge （SOC） of battery pack with traditional method, because SOC is affected by （charging and discharging rate） the rate of charge or discharge, past history, temperature and so on. On the basis of charge and discharge test experiments of battery packs in different operating cases, an artificial neural network simulation model was set up to estimate the SOC of the battery pack, adopted current, voltage and its gradient as input respectively and simulated, realized the estimation of SOC of battery pack. By comparing simulation results with experimental results, simulation results are verified by experimental results, which can proves the correctness of the simulation method.