Electric Vehicle Energy Consumption Estimation with Consideration of Longitudinal Slip Ratio and Machine-Learning-Based Powertrain Efficiency

Abstract

Electric vehicles (EVs) are considered one of the most promising ways to reduce greenhouse gas (GHG) emissions and address fossil fuel shortage. However, due to EV's limited driving range and battery's fluctuating capacity at different temperatures, EV drivers may question EVs’ ability to reach their destinations. In light of this, an accurate EV energy consumption estimation/prediction is vital to relieve drivers' concerns. This paper proposes an EV energy consumption estimation framework that explicitly considers the vehicle longitudinal wheel slip ratio. Besides, a machine-learning-based dynamic efficiency map is devised to capture the energy transfer ratio between electric motor and battery. Furthermore, a mixed second-order L1/H2 estimator is used to calculate the derivatives of velocity data. The method is evaluated based on on-road EV test data, and the result testifies its enhanced performance over a baseline method.