Data-based health indicator extraction for battery SOH estimation via deep learning

Accurately estimating the battery State of Health (SOH) is critical for the safe and stable operation of electric vehicles. In this paper, a data-based method for SOH estimation based on health indicator extraction is proposed. Based on the analysis of the battery aging mechanism and the changing law of the curves, a total of 61 health indicators related to performance degradation have been extracted from the cyclic curves. Discrete Wavelet Transform (DWT) is applied to denoise the raw indicators during preprocessing. A mutual information (MI) feature selection algorithm based on the normalization of the maximum relevance and minimum common redundancy is introduced to select an optimized set of health indicators. Finally, a bidirectional long short memory neural network (Bi-LSTM) model is then used to establish the correlation between the selected health indicators and the battery capacity. The performance of the proposed method is tested by comparing it with other related time series algorithms. The mean absolute error (MAE) of the proposed method is <0.3535 % and the mean absolute percentage error (MAPE) is <0.3747 %, showing that the proposed approach can provide useful guidance for monitoring the safety and stability of battery operation and future research associated to the construction of battery management system.