Interpretable Data-Driven Learning with Fast Ultrasonic Detection for Battery Health Estimation
Kailong Liu, Yuhang Liu, Qiao Peng, Naxin Cui, Chenghui Zhang
Abstract
Dear Editor, Health management is essential to ensure battery performance and safety, while data-driven learning system is a promising solution to enable efficient state of health (SoH) estimation of lithium-ion (Li-ion) batteries. However, the time-consuming signal data acquisition and the lack of interpretability of model still hinder its efficient deployment. Motivated by this, this letter proposes a novel and interpretable data-driven learning strategy through combining the benefits of explainable AI and non-destructive ultrasonic detection for battery SoH estimation. Specifically, after equipping battery with advanced ultrasonic sensor to promise fast real-time ultrasonic signal measurement, an interpretable data-driven learning strategy named generalized additive neural decision ensemble (GANDE) is designed to rapidly estimate battery SoH and explain the effects of the involved ultrasonic features of interest. Specifically, GANDE is derived from improved generalized additive model (GAM) and neural oblivious decision ensemble (NODE), with improved GAM enabling clear interpretability, while NODE provides superior estimation performance. Results illustrate that the GANDE presents satisfactory battery SoH estimation results with R2 over 0.97. In addition, the effects of four ultrasonic features on SoH estimation results can be well explained and visualized, which greatly benefits the understanding of the model's decision making and the control and optimization of battery systems.