Lifetime prediction based on opitimal Loess smoothing and UKF for lithium-ion batteries

Life prediction is the key part of the Prognostics and Health Management for lithium-ion batteries and it is an important means to master the decline tendency of power performance. In this paper, we have established a simplified and effective double exponential distribution capacity attenuation model through the analysis of data from the NASA Ames Center accelerated life test. Based on the model, an infusion method with unscented Kalman filtering algorithm and opitimal Loess smoothing is proposed to predict the life of the lithium-ion battery. The research results show that the presented methods of lithium-ion battery life prediction can effectively estimate the useful life of lithium-ion batteries, thus improving the stability and accuracy of the prediction.