基于鲸鱼算法优化深度极限学习机的锂离子电池剩余使用寿命间接预测

Indirect RUL Prediction of Lithium-ion Battery Based on WOA-DELM

鉴于对锂离子电池直接预测剩余使用寿命(RUL)困难,而极限学习机预测效果不稳定的现状,提出基于等压降放电时间和深度极限学习机(DELM)相结合的间接预测方法。首先,在恒流放电过程中提取出表征电池性能退化的等压降放电时间,分析它与容量间的相关程度并选之作为间接健康因子;其次,引入鲸鱼优化算法(WOA)优化深度极限学习机模型参数,构建锂离子电池RUL预测模型。用锂离子电池数据集中的B0005、B0007两个电池进行实验,结果表明:基于等压降放电时间的WOA-DELM模型预测方法相较于BP神经网络、DELM和PSO-DELM,能够更加准确地预测出锂离子电池的RUL,预测误差±5%,具有较好的预测精度和较快的收敛速度。

In view of the unstable prediction effect of limit learning machine and the difficulty in directly predicting RUL of lithium-ion battery,an indirect prediction method based on constant voltage drop discharge time and deep limit learning machine was proposed.Firstly,in the process of constant current discharge,the equal voltage drop discharge time,which represents the degradation of battery performance,was extracted,and the correlation between it and capacity was analyzed,and it’s selected as an indirect health factor;secondly,the whale optimization algorithm was introduced to optimize parameters of the deep limit learning machine model,and the RUL prediction model of lithium-ion battery was constructed.In order to verify the effectiveness of the method,experiments were carried out on B0005 and B0007 batteries in the lithium-ion battery data set.The experimental results show that,compared with BP method,DELM method and PSO-DELM method,WOA-DELM model prediction method based on equal voltage drop discharge time can predict lithium-ion battery RUL more accurately,and the prediction error range is controlled within±5%.It has better prediction accuracy and faster convergence speed.