基于BP神经网络的电池SOC估算

Estimation of battery SOC based on BP neural network

针对传统动力电池的SOC估计方法的不足,通过编写Matlab程序建立了基于Levenberg-Marquardt(LM)算法的BP神经网络,对其进行了训练及检验。用所建神经网络模型对电池剩余电量进行预测,最大误差小于0.1%。结果满足精度要求,从而验证了所建BP神经网络能够有效地预测蓄电池电压、电流、温度和SOC之间的映射关系。对提高动力电池的能量效率,延长电池的使用寿命具有重要意义。实验表明,此方法提高了电池SOC计算的精度,达到了井下移动救生舱的应用要求。

Due to the deficiencies of the traditional power battery SOC estimation method, Matlab programs based on the levenberg-marquardt （LM） algorithm of BP neural network was wrote and its training and testing was done. A neural network model was built to predict the remaining capacity of the battery. The maximum error is less than 0.1%. Results meet the accuracy requirements, which verify that the BP neural network can effectively predict the mapping between battery voltage, current, temperature and SOC. It has great significance to improve the energy efficiency of the battery power and extend battery life. The experiments show that this method improves the calculation accuracy of the battery SOC to achieve the application requirements of the underground mobile rescue chamber.