State of health estimation for lithium-ion batteries with dynamic time warping and deep kernel learning model

Pengchao Hu, Guijun Ma, Yong Zhang, Cheng Cheng, Beitong Zhou, Ye Yuan

Abstract: The state-of-health (SOH) is an important indicator in battery management system. In order to accurately estimate the SOH of lithium-ion batteries, an improved Gaussian process regression (GPR) model named deep kernel learning (DKL), combining with dynamic time warping method is proposed in this paper. Use of voltage curve during constant current charing procedure, two features are extracted as the inputs of the DKL model. Particularly, an important feature is extracted by dynamic time warping which makes full use of the information in the voltage curve to avoid a subjective feature extraction. DKL is used for SOH modeling, which replaces the traditional kernel function of GPR model with neural networks. A battery experiment at room temperature is designed to validate the performance of the proposed method. The results show that the proposed method has good performance using small training samples.