Research On Remaining Useful Life Prediction Method Of Lithium-ion Battery Based On Wiener Process

Lithium-ion batteries are widely used in industrial,military,and social fields due to their high-density energy and long service life.However,safety accidents caused by their failure often occur.Accurately predicting the remaining useful life(RUL)of their are significant to guarantee the reliability and economy of the battery operation process.Because Wiener process can describe non-monotonic degradation process,and has good physical interpretation and mathematical characteristics,it has become one of the most mainstream models in RUL.However,due to the change of operating conditions and internal mechanisms,the degradation of lithium-ion battery has time-varying uncertainty,change point,or capacity recovery issues,resulting in the two-phase or even multi-phase phenomenon in the degradation trajectory,where using single-phase Wiener process to apply to the above problems often inadequate.Therefore,this paper starts with the analysis of lithium-ion battery degradation.Aiming at the problems of degradation uncertainty,change point,and capacity recovery,the RUL prediction methods of lithiumion battery based on Wiener process are studied from the "single-phase to two-phase to multi-phase" idea.The main research contents of this paper are summarized as follows:Firstly,the composition and working principle of the lithium-ion battery are introduced,and the cycle lifetime is determined as a performance index for measuring the battery life.Then the NASA and CALCE experimental dataset are visualized and the influencing factors of battery degradation are analyzed.Combined with experimental data,the problems of change point and capacity recovery during the degradation process of lithium-ion battery are further confirmed.Secondly,aiming at the degradation uncertainty of lithium-ion battery,and according to the characteristics and lifetime definition of the single-phase Wiener process model,the RUL prediction method for lithium-ion battery based on the linear Wiener process is given.The maximum likelihood estimation(MLE)method is adopted to identify the unknown parameters of proposed model.Based on this,the differences and nonlinearity among individuals are further considered,and the probability density function(PDF)of life is derived.Finally,the effectiveness of the method is verified by NASA experimental data.Then,aiming at the change point in the degradation process of lithium-ion battery,a two-phase Wiener process degradation model is established.Taking into account the randomness of the initial state transition to the changing state and the variability between unit degradations,the analytical solution of RUL under the concept of the first passage time(FPT)is derived.The EM algorithm and Bayesian method are used to identify the model parameters.Finally,the rationality of the method is verified by numerical simulation and CALCE data.Finally,aiming at the problem of capacity recovery during the degradation of lithium-ion battery,the multi-phase degradation model with random jumps is established with a discontinuous drift function.Then fully considering the randomness of the capacity mutation value and the uncertainty of the amount of degradation at the restoration state,the FPT distribution is derived.The two-phase empirical MLE method is used to estimate the capacity recovery information and the degradation model parameters.Finally,the validity of the method is verified by numerical simulation and NASA experimental data.