Remaining Useful Life Prediction of Lithium-ion Battery Based on Discrete Wavelet Transform

Abstract

Abstract The remaining useful life (RUL) and state-of-health (SoH) are critical to battery management system (BMS) to ensure the safety and reliability of electric vehicle (EV) operation. Recent literatures have reported various methods to estimate the RUL and SoH with a focus on the capacity loss, internal resistance increase, voltage drop, self-discharge, number of cycles, etc. However, most of the works only consider the battery in certain cases without thinking about the factors in real operation. In some cell internal parameters identification approaches, the accuracy of prediction results highly depends on the model and identification method. Aiming at these problems, a model free RUL prediction method is proposed in this work based on the discrete wavelet transform (DWT). The dynamic stress test (DST) schedule is conducted on the commercial 1665130-type lithium-ion battery and various outputs with non-stationary and transient phenomena are obtained and analyzed. Finally, experiments are conducted on the lithium-ion battery with different aging levels to verify the proposed method, and the results indicate that high accuracy of RUL prediction can be obtained by this quantitative correlation.