Research On Remaining Useful Life Of Lithium-ion Batteries With Particle Filter Algorithm

Lithium-ion batteries,as an important source of energy for portable electronic devices and electric vehicles,can cause serious accidents and economic losses,and even threaten the lives of people.If the aging rate of lithium-ion batteries can be accurately known during the recycling process,the batteries can be replaced in time before the failure occurs,thereby effectively avoiding accidents and ensuring the normal operation of the entire system.However,since the aging rate of the batteries cannot be accurately measured,it is necessary to study the prediction of the remaining useful life(RUL).This article takes the 18650-type cylindrical lithium-ion batteries as the research object,and studies the remaining useful life of the lithium-ion batteries in recycling use.It is mainly divided into the following aspects:Firstly,this paper theoretically analyzes the working principle of lithium-ion batteries and the mechanism of its lifetime degradation,selects the battery capacity as an indicator to evaluate the end of battery life,and introduces the mathematical basis and implementation steps of the particle filter algorithm.At the same time,the double exponential model and polynomial model are compared and analyzed systematically based on experimental data analysis,and the double exponential model with good matching of batteries data is selected as the focus of this study.Then we calculate the parameters of the lithium-ion battery capacity degradation model and predict the remaining useful life of Lithium-ion batteries based on particle filter algorithm.Secondly,the factors affecting the prediction accuracy of remaining useful life of Lithium-ion batteries are analyzed.As to the particle deprivation in resampling process of particle filter,an improved unscented particle filter algorithm based on Monte Carlo Markov chain(MCMC)is proposed.That is,a better importance density function is selected by the unscented particle filter,and after the step of resampling,The MCMC algorithm is introduced to generate new particles to ensure the diversity of particles,thereby the problem of particle degradation in the basic particle filter algorithm can be solved more comprehensively.The simulation experiment show that the improved model can achieve higher precision than the fundamental particle filter algorithm,the validity and availability of our model was verified.Finally,As to the commonly used capacity degradation model can not accurately reflect the local characteristics of capacity degradation to ensure prediction accuracy of lithium-ion batteries,three capacity degradation models are proposed by combining the double exponential model with the polynomial model based on experimental data analysis.Then,through the analysis of adjusted R square,AIC criterion,root-mean-square and other indicators,an improved capacity degradation model with better matching degree is established.The simulation experiment verify the predictive performance and applicability of the combined model.It provides a way to improve the capacity degradation model.