Data Based Investigation of Battery Electric Vehicle Performance and Battery Capacity Prediction Using Machine Learning

Battery electric vehicles (BEV) are becoming widespread to fulfill zero tailpipe carbon-dioxide (CO2) and gaseous tailpipe emissions, which are enforced by recent regulations. For current BEVs, range is one of the big challenges since their range is comparably shorter than that of internal combustion engine vehicles (ICEV). To have an acceptable driving range, it is important to identify the amount of energy required from the battery. In this study, open-source available BEV data is collected to investigate the performance, range and battery capacity requirements based on available data and to understand current level of BEV technology. Collected data contains more than 75 different BEV models from various car manufacturers and different models. These cars are passenger cars like saloon cars, hatchbacks, and SUVs, so commercial vehicles like vans, trucks, etc. are not included. Collected data is used to build different machine learning (ML) models to predict the required battery capacity for a target range based on the Worldwide Harmonized Light-Duty Vehicles Test Procedure (WLTP) cycle.