A Degradation Model Based On Jump Diffusion Process And Its Application In Lithium-ion Battery Data

Based on the recognition of the research status of degradation models,this paper takes lithium battery as an example to study the product degradation process with selfrecovery phenomenon.As the main on-board power supply of electric vehicles,lithium batteries have developed rapidly in recent years.At the same time,the safety and life of lithium-ion batteries has received widespread attention,and all these problems have a considerable relationship with the battery performance change.By modeling and studying the performance degradation process,the remaining service life of the system can be predicted accurately,and the system can be maintained or replaced in advance before the failure,so as to ensure the safe and reliable operation of the system.In the process of research on lithium battery data,this paper uses the jump diffusion model widely used in the economic field.And according to the distribution characteristics that the change rate of battery capacity shows a right skewed peak,a single exponential jump diffusion model is proposed to describe the degradation path of products with self-recovery phenomenon.For the estimation of model parameters,this paper adds Markov chain Monte Carlo method(MCMC)on the basis of LM jump identification method,and combines the advantages of the two methods for different parameter estimation,which presents a combined algorithm to estimate the parameters more accurately.In order to compare the two parameter estimation methods,jump identified-MCMC combination algorithm and LM jump identification,Monte Carlo simulation experiment is conducted in this paper.From the perspective of related indexes such as root mean square error,experimental results show that the parameter estimation results obtained by using the combined algorithm are closer to the real parameter values,and the combined algorithm is more stable and accurate than the LM jump identification method.Furthermore,the residual useful life density distribution function of the product is approximate estimated by Monte Carlo method.It can be seen from the distribution function images and the corresponding JS divergence values that the distribution function corresponding to the combined algorithm is closer to the distribution function corresponding to the real parameters.At the same time,the combined algorithm has a smaller average absolute percentage error in predicting the average remaining useful life.These results show that the combined algorithm has a good feasibility in predicting the remaining useful life.Finally,this paper conducts an empirical study on the real capacity data of lithium batteries,and draws the same conclusion as the simulation experiment.Based on the approximate distribution function of the remaining service life of lithium batteries,relevant suggestions are put forward from the perspective of safety.