Learning-based control strategies for hybrid electric vehicles

Hybrid electric vehicles use control strategies to distribute the torque requested by the driver between the internal combustion engine and the electrical motor. Many different types of control strategies have been proposed, but in general it is impossible to determine which control strategy performs best if the future driving conditions are unknown. In this paper, we introduce two learning-based control strategies which use an arbitrary set of basic control strategies in order to minimize the fuel consumption of the hybrid electric vehicle. Our simulation results show that the fuel consumption of the learning-based control strategies are comparable to the fuel consumption of the best basic control strategy in the set even without a priori knowledge of the driving conditions.