Explainable Data Driven Digital-Twins For Predicting Battery Status In Electric Vehicles Using Machine Learning

Abstract

Electric Vehicles (EVs) are emerging as a cornerstone of sustainable transportation, where battery systems play a critical role in determining performance, safety, and operational lifespan. Conventional Battery Management Systems (BMS) primarily focus on real-time monitoring of parameters such as voltage, current, and temperature, offering limited predictive capability and minimal interpretability. This study proposes an explainable, data-driven digital twin framework for predicting EV battery behavior using machine learning techniques. The framework constructs a virtual representation of the battery system that dynamically simulates performance based on input variables including speed, State of Charge (SOC), State of Health (SOH), temperature, terrain, and auxiliary loads. Machine learning models are employed to estimate key performance metrics such as driving range, energy consumption, and battery degradation trends. To enhance transparency, Explainable Artificial Intelligence (XAI) techniques are integrated, enabling interpretation of model predictions and identification of influential parameters. The system is implemented using a FastAPI-based backend for simulation and prediction, coupled with a Next.js interactive dashboard for visualization. The proposed approach facilitates predictive analytics, scenario simulation, and informed decision-making, thereby extending traditional BMS capabilities toward a more intelligent and user-centric EV battery management solution.

DOI: https://doi-ds.org/doilink/04.2026-65651221