Explainable data-driven digital twins for predicting battery states in electric vehicles

Battery state prediction is crucial for automobile industry transition to electric vehicles (EVs) in order to opt imize performance, safety, and longevity. In this paper an Explainable data-driven digital twins are used to anticipate EV battery conditions in a novel way. Advanced machine learning algorithms have been integrated including Radial Basis Function (RBF) networks, Random Forests (RF), Feed-forward Neural Networks (FNN), Conventional Neural Networks (CNN), Long Short-Term Memory (LSTM), Radial Neural Networks (DNN), Support Vector Regression (SVR), Support Vector Machines (SVM), and Extreme Gradient Boosting (XGBoost). The main goal of developing battery management systems for electric mobility is to improve battery state predictability and provide accurate predictions of the battery health and charge levels under various circumstances.