Study On Estimation Method Of Remaining Useful Life Of Lithium-ion Battery

Today,there are many bottlenecks in the development of new energy vehicle technology,mainly due to issues such as battery cruising ability,safety,and reliability.Therefore,research on the performance of electric batteries is of great importance.Power lithium-ion batteries are regarded as the end of Life when their Remaining capacity reaches 70%～80%of the rated capacity.In order to maintain and replace defective batteries in a timely manner,it is necessary to predict the Remaining Useful Life(RUL)of lithium-ion batteries.This document studies the factors that affect the degradation of single lithium-ion batteries,combined with the empirical model of battery capacity degradation and the extended H∞ particle filter algorithm,this article proposes an efficient method to estimate the RUL of the battery.The following aspects is the main research in this work:(1)Select the empirical model of battery capacity degradation based on the data from the battery cycle aging experiment.In this article,the 18650 cylindrical LiFePO4 battery is the subject of investigation,and the RUL of the lithium ion battery is studied before reaching the failure threshold,and the failure threshold is set to 80%.Perform cyclic charge and discharge experiments on individual batteries at different discharge rates.Based on the corresponding relationship between battery capacity and cycle times,an empirical model of battery capacity decline is established.The Matlab fitting tool is used to compare and analyze the degree of curve fit and root mean square error of different fitting models,the double exponential model is selected as the basic empirical battery model for this study.(2)Battery RUL prediction based on particle filter(PF)and its improved algorithm.The parameters of the battery dual exponential model were taken as the initial state parameters of PF algorithm,and the residual capacity was taken as the observation.The battery state parameters were updated in real time based on the particle filter algorithm to realize the online prediction of the battery RUL.In order to solve the particle degradation problem existing in the standard PF algorithm,the extended filter algorithm was used to generate the importance probability density function,and each particle was updated during the importance sampling process.Based on the extended H∞ particle filter algorithm,the prediction of RUL for lithium-ion batteries was realized.Compared with the standard particle filter algorithm,it is found that the proposed improved particle filter algorithm has higher prediction accuracy.It can also effectively solve the particle degradation problem in the PF algorithm.(3)Adaptability verification of extended H∞ particle filtering algorithm.The extended H∞particle filter algorithm was validated based on the data of four groups of lithium cobalt oxide batteries published by NASA.It is found that the improved particle filter algorithm can realize the RUL prediction of lithium-ion batteries of different materials with small errors,which improved the effectiveness of the algorithm proposed in this work.