딥러닝 및 XAI 기법을 활용한 리튬 이온 배터리의 잔여 수명 예측

초록

This paper proposes a data-driven method to predict State-of-Health (SoH) and Remaining Useful Life (RUL) of Li-Ion battery used in Electric Vehicle (EV) based on Deep Learning and XAI (Explainable AI). Battery life prediction algorithm is designed using various Deep Learning models based on Transfer Learning, and the algorithm is verified through the Lithium-Ion battery dataset provided by NASA Ames. In addition, Deep Learning model analysis is performed using SHAP (Shapley Additive exPlanations), a technique of XAI, as a solution to the lack of interpretation of the prediction results, which is a problem when using a data-driven method for predicting the RUL. Through this, the optimal Deep Learning model for predicting SoH and RUL of Lithium-Ion battery and changes in battery life prediction performance according to the number and kinds of feature factors affecting the prediction result are analyzed.

키워드