Research Of Remaining Useful Life Prediction Method And Degradation State Identification On Lithium-ion Battery

Since the implementation of commercialization of lithium-ion battery in 1990 s,lithium-ion battery become the main storage of next generation mobile communication for equipment and other electronic products with its high voltage,high specific energy,long life,low self discharge instead of the Ni Cd and Ni MH batteries,and the power lithium-ion battery is gradually expand its application in the field of aerospace.However,the battery system is one of the most vulnerable parts of the satellite.Once the power system fails,it will have a significant impact on the overall operation of the satellite.Therefore,the state identification and performance prediction of lithium-ion batteries have become one of the most hot topics in the space power system.This paper is focused on the degradation state identification and remaining useful life prediction of lithium-ion batteries.The main work is as followsFirstly,the working principle and performance degradation analysis of lithium ion batteries are introduced,then lists the commonly used lithium-ion battery equivalent circuit model and model parameter identification method,simulate lithium-ion battery fade modeling by COMSOL multiphysics simulation software,The capacity fade mechanism of lithium-ion batteries is further researched by simulation,the simulation data of the battery life are used to predict the remaining life of the lithium-ion battery.Model contains the performance of lithium-ion battery at different discharge current decline trend.Secondly,a state recognition method of lithium-ion battery degradation based on the study and analysis of recurrent neural network and battery health index is put forward,a kind of SOH estimation of discharge time series based on equal voltage intervals is proposed through comparison and analysis of several health index.Finally,The feasibility of lithium-ion battery RUL prediction based on ARMA and PF is verified by using NASA PCoE public data set and the simulation data,a comparison is carried out in these two methods.The method mentioned in th is paper can solve online lithium-ion battery health status prediction by combining the degradation state recognition method because of data monitoring conditions limited and less monitoring data.