Battery power management and effective maintenance contribute a major part to electric vehicle’s performance. Particularly in all-terrain vehicles like SAE E-BAJA, the car has to drive to an optimal range in off-road conditions. To ensure the vehicle can operate at an optimal range under off-road conditions, a predictive analysis was conducted using Machine Learning for various transmission ratios using the algorithm known as Random Forest. This approach was employed to measure the Battery's State of Charge (SoC) and power consumption across different transmission ratios. Before applying machine learning, the vehicle underwent extensive real-time testing on an off-road track. During these tests, the State of Charge of the battery, indicated by voltage levels, was carefully monitored. The battery's peak voltage was recorded at 51.9 V, with a minimum acceptable voltage of 49.5 V. Tests were conducted at various transmission ratios, including 6.136, 7.771 and 8.590 under consistent track and environmental conditions to ensure reliable data. The real-time data gathered from these tests, such as SoC and power consumption at different transmission ratios, were then fed into the Machine Learning model. The Random Forest algorithm processed this data to predict performance metrics, such as battery SoC, power consumption, and torque, for each transmission ratio. The predictions from the Machine Learning model are validated against the real-time test results to ensure accuracy. This study provides valuable insights into how transmission ratios impact the battery’s performance, which ultimately impacts the vehicle range for various transmission ratios in off-road conditions.
Keywords: Transmission ratio, Battery discharge characteristics, Random Forest algorithm, Python, Machine Learning, All-Terrain Vehicle.