Real-Time Mixed-Integer Energy Management Strategy for Multi-Motor Electric Vehicles

Paper i proceeding, 2023

This paper11This work was supported by the Sweden's Innovation Agency (Vinnova) under Grant 2017-05506. presents a real-time capable energy management strategy for multi-motor electric vehicles, based on mixed-integer model predictive control (MI-MPC). In this strategy, torque allocation and clutch on-off are co-optimized to minimize both the energy consumption and the frequent changes in clutch engagement status. To be able to solve the mixed-integer (MI) problem in real time, we propose a bi-level programming approach in which the torque allocation subproblem is solved at the inner level using an explicit closed-form analytical solution, while the integer decisions are optimized at the outer level using implicit dynamic programming (i-DP). The simulation results show that the proposed strategy can achieve up to 11 % energy savings, depending on the load demand in a driving mission, compared to a rule-based controller typically used in production vehicles. In addition, the proposed approach is guaranteed to find the global optimum for the MI problem in each MPC update. With a mean time to solution of around 4.6 ms, the proposed strategy shows promising real-time capabilities for online implementation in multi -motor electric vehicles.