Predictive Analysis of Electric Vehicle Performance Parameters Using Random Forest Regression: A Comprehensive Study of Battery Capacity, Range, Charging Time, And

Abstract

This study presents a comprehensive analysis of electric vehicle (EV) technology, focusing on heavy-duty electric automobiles weighing more over 10,000 pounds gross vehicle weight. The research examines advanced electric powertrains with dual electric drive units for the front and rear axles, each incorporating electric motors, controllers, two-speed gearboxes, and locking differentials. The “high-torque avoidance control” system is highlighted, which dynamically adjusts transmission gear ratios to optimize motor rotation speed and maintain vehicle efficiency, while avoiding inefficient operating regions characterized by PWM over-modulation control. The study evaluates critical performance parameters including battery capacity (kWh), driving range (km), charging time (hrs), and maximum speed (kmph) across various EV configurations. The research addresses the evolution from traditional internal combustion engines to sustainable electric alternatives driven by environmental concerns, petroleum resource limitations, and rising fuel costs. Extended-range electric vehicles (EREVs) have been identified as promising solutions to address battery electric vehicle limitations. This method uses random forest regression for predictive modelling, using ensemble learning to analyse complex, nonlinear relationships between vehicle parameters. This approach provides robust predictions while managing high-dimensional datasets and reveals feature importance, making it particularly suitable for vehicle performance analysis and optimization of electric vehicle systems.