Enhancing Model Robustness in Battery Health Prediction through Advanced Input Features

In this study, a new AI-based model has been proposed for electric vehicle battery state of health estimation. Our approach relies on the use of a Convolutional Neural Network (CNN), which takes into account not only voltage, current, and state of charge (SoC), but also other characteristics such as ampere-hours and their derivatives, as well as SoC and its derivative. We perform a comparison study between two models that use different datasets for training. This comparative analysis aims to assess the effect of different input variables on the learning model’s accuracy in predicting battery health.