基于改进粒子滤波算法实现锂离子电池RUL预测

In view of the problem that the prediction accuracy of the remaining useful life (RUL) of lithium-ion batteries is low due to the loss of particle diversity in the process of using the traditional particle filter algorithm to predict the RUL, the idea of linear optimization resampling is introduced, and a battery RUL prediction method based on linear optimization resampling particle filter (LORPF) is established. This method uses a double exponential model as the battery aging model, iteratively updates the model parameters through the LORPF algorithm, realizes the prediction of the battery RUL and gives the uncertainty expression of the prediction result. Finally, the battery data from the NASA PCoE Research Center and the battery data from an independently built experimental platform are used to compare and verify the proposed method with the traditional PF method. The results show that this method effectively improves the RUL prediction accuracy, and its error is less than 5%.

鉴于采用传统粒子滤波算法来预测锂离子电池剩余使用寿命(RUL)过程中,存在粒子多样性丧失现象而导致RUL预测精度较低的问题,引入线性优化重采样思想,建立了基于线性优化重采样粒子滤波(LORPF)的电池RUL预测方法。该方法以双指数模型作为电池老化模型,通过LORPF算法对模型参数进行迭代更新,实现电池RUL预测并给出预测结果的不确定性表达,最后使用美国国家航空航天局PCoE研究中心的电池数据和自主搭建实验平台的电池数据对所提方法与传统PF方法进行对比验证,结果表明该方法有效提高了RUL预测精度,其误差小于5%。