Predicting Electric Vehicle Energy Consumption from Field Data Using Machine Learning

Abstract

Accurate prediction of electric vehicle (EV)
energy consumption is essential for range estimation, battery
management, and intelligent energy optimization. This paper
presents a machine learning–based approach for predicting
EV energy consumption using real-world field data. The
dataset incorporates key operational and environmental
parameters, including vehicle speed, acceleration patterns,
driving behavior, road conditions, ambient temperature, and
auxiliary load usage. After data preprocessing and feature
engineering, multiple supervised learning models—such as
Linear Regression, Random Forest, and Gradient Boosting—
are developed and evaluated. Model performance is assessed
using standard metrics including Mean Absolute Error (MAE),
Root Mean Square Error (RMSE), and R² score. Experimental
results demonstrate that ensemble-based models outperform
traditional regression techniques in capturing nonlinear
relationships within field data. The proposed framework
improves prediction accuracy and provides a scalable solution
for real-time energy estimation in EV systems. This work
contributes to enhancing range reliability, energy efficiency
optimization, and intelligent transportation system
development.