Research On Prediction Of Remaining Useful Life Of Lithium-ion Battery Based On Improved Particle Filter

The electrochemical characteristics of Li-ion batteries are complex.With continuous charge and discharge cycles,its usable capacity will decline,which called capacity degradation.When the capacity decline to 70% ~ 80% of the rated capacity,it can be regarded as battery failure.Usually,Kalman filter and its improved algorithms,particle filter and its improved algorithms are used to predict the remaining useful life of the battery.Inspired by the characteristics of particle filter,an improved particle filter algorithm is proposed in this paper,which named as Exponential Smoothing Particle Filter algorithm.The work and achievements of this paper are mainly included in the following aspects:(1)The electrochemical reaction process,degradation mechanism and the main factors that affect the life of lithium-ion battery are analyzed.The tests of lithium-ion battery charge and discharge cycles are designed and completed.The capacity degradation curves of lithium-ion batteries,which are used to support the follow-up prediction research work,are obtained by processing the test data.(2)After that,Extended Kalman filter and residual resampling particle filter are used to fully track the capacity degradation curve of lithium-ion battery;extended Kalman filter,residual resampling particle filter and exponential smoothing particle filter are used to predict the remaining useful life of lithium-ion battery.The prediction results and the validity of the proposed method are verified by the data from the charge and discharge cycle test of lithium-ion battery.Absolute error,relative error and stability error are introduced as the evaluation indexes of prediction results.The three filtering methods in this paper are compared and analyzed comprehensively.(3)The results show that: The tracking ability of extended Kalman filter is weaker than that of residual resampling particle filter;The same rule of the three filtering methods is that the prediction accuracy and stability are improved with the start point of prediction moving backward;The prediction accuracy and stability of the proposed exponential smoothing particle filter are the best among the three methods,followed by the residual resampling particle filter,and the prediction accuracy and stability of the extended Kalman filter are weaker than the former two.The research in this paper improves the accuracy and stability of the filter method for the prediction of the remaining useful life of lithium-ion batteries,and optimizes the historical parameter processing ability of the particle filter.It is believed that this paper can help to improve and optimize the particle filter algorithm,and it is useful in the field of battery life research,and it has a positive impact on the application and development of battery management technology.