基于改进Bayes信息量准则的锂电池自适应变阶AVO模型

Adaptive variable-order model of lithium ion battery based on improved Bayes information criterion

提出了一种融合Bayes信息量准则和樽海鞘优化算法的自适应变阶(AVO)模型。该模型以端电压误差的Bayes信息量作为一阶和二阶RC模型的变阶依据,利用樽海鞘优化算法(SSA)搜索Bayes信息量序列的全局最优解,实现对模型最优变阶序列的求解;通过开展混合动力脉冲特性(HPPC)实验,选用含遗忘因子递推最小二乘法(FFRLS)获取精确的模型参数,完成AVO模型的电阻及电容参数辨识和精度在线验证。结果表明:所建立的锂电池自适应变阶AVO模型平均误差为0.011 9 V,精度及实用性较高。

An adaptive variable-order(AVO)model that combines Bayes information criterion(BIC)and salp swarm algorithm(SSA)was proposed.The model took the Bayes information of terminal voltage error as the order change basis of the first-order RC model and the second-order RC model,and used the SSA to search the global optimal solution of the Bayes information sequence,so as to solve the optimal order change sequence of the model.Through the hybrid pulse power characterization(HPPC)experiments,the recursive least squares algorithm with forgetting factor(FFRLS)was selected to obtain accurate model parameters,and the resistance and capacitance parameters identification and accuracy online verification of the AVO model were completed.The results show that the average error of the established lithium ion battery AVO model is 0.0119 V,which is highly accuracy and practical.