Adaptive model predictive control for hybrid electric vehicles power management

For hybrid electric vehicles (HEVs) that have predictable routes, it is beneficial to use model predictive control (MPC) to derive the optimal control trajectories that can minimize fuel consumption while achieving other objectives. However, for applications such as city buses and delivery trucks, even though the driving routes are pre-determined, the loads of the vehicles are changing from time to time. The trajectories computed by the dynamic programming (DP) or other optimization algorithms based on the pre-defined model might not be optimal during real-time operation. Therefore, an adaptive control design with on-line vehicle parameter estimator is needed to account for those unpredictable changes. In this work, we propose an adaptive model predictive control (AMPC) design that can estimate and update the vehicle mass in real-time. A comparative case study is conducted to analyze the effectiveness of adaptation by comparing the AMPC and non-adaptive MPC in terms of fuel economy. An MPC algorithm based on DP is integrated with a parameter estimation algorithm based on the least squares, simulation results based on a comprehensive vehicle model is presented in this paper.