A Hybrid Approach Based on Gaussian Process Regression and LSTM for Remaining Useful Life Prediction of Lithium-ion Batteries

Abstract

Accurate remaining useful life (RUL) prediction is of great importance to the battery management second-life utilization. This paper proposes a novel hybrid data-driven RUL prediction method based on Gaussian process regression (GPR) and long-short term memory neural network (LSTM). An initial prediction of RUL through LSTM is employed as the mean function of GPR instead of simply assuming it to be zero or a linear form. The aging data of four batteries from NASA data repository is used for model verification and comparison. The results show that the proposed LSTM-GPR approach has higher prediction accuracy than the traditional LSTM and GPR approaches with less training data.