Prediction Of Remaining Useful Life For Lithium-ion Batteries Considering Capacity Regeneration

In this paper,the remaining useful life(RUL)prediction of lithium-ion batteries is investigated based on the capacity regeneration phenomenon.The main research contents and results are as follows:(1)A model based on variational modal optimization decomposition and correlation vector machine is constructed for the error problem caused by the capacity regeneration phenomenon of lithium-ion batteries.The multi-scale decomposition of lithium-ion battery capacity is performed by using the variational modal optimization decomposition,and then the least squares algorithm and the correlation vector machine algorithm are used to train and predict each sequence to obtain the RUL prediction results of lithiumion batteries.The errors of root mean square error,mean absolute error and mean absolute percentage error obtained from the prediction are within 1%,indicating that the model in this paper can effectively improve the prediction accuracy of the remaining service life of lithium-ion batteries.(2)To address the problem that the parameter space of variational modal optimal decomposition needs empirical setting,a model based on adaptive optimal decomposition algorithm and correlation vector machine is constructed.The adaptive optimal decomposition algorithm is used to realize the adaptive noise reduction decomposition of the lithium-ion battery capacity sequence,and then the decomposition sequence is input into the least squares algorithm and the correlation vector machine algorithm for training and prediction to obtain the RUL prediction results of the lithium-ion battery.Compared with the multi-layer perceptron,the model combining empirical modal decomposition with autoregressive integrated moving average,the long and short-term memory recurrent neural network and other common models,the adaptive optimization based decomposition algorithm and the correlation vector machine model proposed in this paper have higher prediction accuracy,and the combined model given in this paper performs optimally under the conditions of setting different prediction starting points.(3)For the problem that the capacity regeneration curve obtained from the sequence decomposition of lithium-ion battery is difficult to be fitted by the correlation vector machine algorithm in the data set where the capacity regeneration phenomenon is not obvious,a model based on adaptive optimal decomposition and limit learning machine is constructed.The adaptive optimal decomposition algorithm is used to decompose the lithium-ion battery capacity sequence with noise reduction,and then the decomposition sequence is assigned by Mann-Kendall test and input into the least squares and extreme learning machine algorithms for training to realize the end-to-end RUL connection prediction of the model.The prediction results obtained by the model based on the adaptive optimal decomposition algorithm and the limit learning machine proposed in this paper are highly generalizable and accurate.This paper addresses three problems of RUL prediction for lithium-ion batteries.Three models are designed respectively,which all achieve good prediction accuracy and can be effectively applied to the RUL prediction of lithium-ion batteries under different conditions.