基于PSO-RBF神经网络的锂离子电池健康状态预测

摘要

针对传统方法估计锂离子电池健康状态(SOH)时内部参数测量困难等问题，提出一种基于粒子群优化径向基函数神经网络的锂离子电池 SOH 预测方法.通过对锂离子等效模型的研究，结合充放电过程的实验数据，确定了影响锂离子电池 SOH 特性的几个关键参数.将试验数据输入仿真模型进行网络训练和校验.仿真证明，相比 BP 神经网络和普通 RBF 神经网络，该算法的预测精度可提高20％，节省66．7％以上的优化时间.%For the traditional method to hardly estimate the internal parameters of Li-ion battery SOH,a PSO algorithm based on RBF neural network for SOH prediction of Li-ion batteries was pro-posed.Based on the Li-ion battery equivalent model,several key parameters which affected the SOH characteristics of the battery were determined by experimental data of the charged and discharged processes.The test data were input simulation model for network training and verification.Simulation results show that,compared to the BP neural network and the general RBF neural network,the algo-rithm may increase 20% of prediction accuracy,save more than 66.7% of the optimization time.