Smart Strategies for Improving Electric Vehicle Battery Performance and Efficiency

Area of Science:

• Artificial Intelligence
• Automotive Engineering
• Data Science

Background:

• Growing demand for electric vehicles (EVs) requires accurate range prediction to mitigate range anxiety.
• Optimizing driving parameters like acceleration and velocity is crucial for enhancing EV user experience and efficiency.

Purpose of the Study:

• To develop and evaluate a machine learning (ML) framework for predicting EV range, optimal acceleration, and velocity.
• To compare the performance of individual ML models against ensemble combinations for predictive accuracy.

Main Methods:

• A synthetic dataset of 2,000 real-world driving scenarios was generated.
• Four ML models (Random Forest, Extra Trees, Linear Regression, LSTM) were trained and tested individually and in ensembles.
• Statistical reliability was ensured through ten independent runs with randomized data partitioning.

Main Results:

• Ensemble models consistently outperformed individual models in predicting EV range and driving parameters.
• The full ensemble (RF+ET+LSTM+LR) demonstrated the most robust performance across MAE, MSE, and R² metrics.
• A real-time web application was developed for dynamic estimation of driving parameters.

Conclusions:

• AI-driven predictive modeling offers significant potential for improving EV energy management and driving efficiency.
• The proposed framework can support efficient driving behaviors and reduce range anxiety for EV users.
• Integration of ML into EV systems is key to advancing sustainable transportation.