Embedded Model Predictive Control for Torque Distribution Optimization of Electric Vehicles

Torque distribution optimization of electric vehicles should consider various demands, including energy conservation, power, and tire antislip, which are usually coupled. Furthermore, there is a higher requirement for real-time feasibility of controllers. This study proposes an embedded model predictive controller called EmMPC and applies it to the torque distribution optimization of electric vehicles equipped with two or more motors. In EmMPC, barrier functions and normalized projected gradients are utilized to handle state inequality and linear input constraints. A fast line search method combined with heuristic search regions and quadratic function properties is presented to reduce computational burden further. Subsequently, an offline and online combined torque distribution optimization strategy is presented based on EmMPC and an analysis of maximum energy-saving potential. Finally, hardware-in-the-loop experiments were performed with an embedded vehicle-mounted controller to illustrate the effectiveness of our method.