State-of-health estimation and remaining useful life prediction for lithium-ion batteries based on an improved particle filter algorithm

State-of-health (SOH) and remaining useful life (RUL) are vital indicators closely related to the safety of lithiumion batteries (LIBs). In this study, an online capacity estimation and offline RUL prediction methods based on an improved particle filter and recursive-least-square (PF-RLS) algorithm are proposed. In this method, the characteristic voltage (CV) is extracted from the discharge curve as a health feature, and the correlation model of CVcycles-capacity is established. Then, an improved PF-RLS algorithm is used to estimate the CV in real-time to realize SOH estimation and RUL prediction. In the improved PF-RL algorithm, the initial value of the proposed probability density is optimized by fitting the sample battery aging data to improve the accuracy and rapidity of the model parameter identification. The results show that the prediction accuracy and stability of the improved PF-RLS algorithm are better than those of the standard PF algorithm. The SOH estimation error can be kept within 3 %, and the RUL prediction error can be kept within 5 % during the battery aging process.

Automated summary

In this study, an online capacity estimation and offline RUL prediction method based on an improved particle filter and recursive-least-square (PF-RLS) algorithm is proposed. The characteristic voltage (CV) is extracted from the discharge curve as a health feature, and the correlation model of CV-cycles capacity is established. The results showed that the improved PF-RLS algorithm has better prediction accuracy and stability than the standard PF algorithm, with an SOH estimation error within 3% and an RUL prediction error within 5% during battery aging.