Forecasting the state of health and remaining useful life of Li-ion batteries is an unsolved challenge that limits technologies such as consumer electronics and electric vehicles. This paper presents an accurate battery forecasting system by combining electrochemical impedance spectroscopy (EIS) with Gaussian process machine learning. Over 20,000 EIS spectra of commercial Li-ion batteries are collected, forming the largest dataset of its kind. The Gaussian process model uses the entire spectrum as input, automatically determining which spectral features predict degradation and accurately predicting remaining useful life, even without complete knowledge of past operating conditions.

Yunwei Zhang, Qiaochu Tang, Yao Zhang, Jiabin Wang, Ulrich Stimming, Alpha A. Lee