Remaining useful life prediction of – Lithium batteries based on principal component analysis and improved Gaussian process regression
Jiang Xing, Huilin Zhang, Jianping Zhang
Abstract
There are several problems in the traditional model-based method of predicting the state of health (SOH) and remaining useful life (RUL) of lithium batteries, including complex modeling and low prediction accuracy. The strategy for predicting the RUL of lithium batteries in this study is based on Principal Component Analysis (PCA), the health Indicator (HI), and improved Gaussian process regression (IGPR). First, according to the cycle curve of battery voltage during charging, four parameters were extracted as the health Indicator (HI) of battery, and then Spearman (SCA) was performed to examine the relationship between the HI and capacity of the battery. Finally, the RUL was predicted using the SVR, GPR and PCA-IGPR networks, and the indices were compared and analyzed. The experimental outcomes demonstrate that the suggested RUL prediction model based on PCA fused with HI and IGPR has small error, strong robustness, and good application prospects.