Non-linear model predictive control of a power-split hybrid electric vehicle

Hoseinali Borhan, Ardalan Vahidi (Clemson University), Wei Liang, Anthony Phillips, Stefano Di Cairano, Ming Kuang, Ryan McGee (Ford)

Power-split HEV configuration

This paper builds on the previous published works in which the authors had employed nonlinear model predictive control for the (sub)optimal power management of a power-split hybrid electric vehicle (HEV). In addition to the battery’s state of charge, this work includes the effect of inertial powertrain dynamics in the control-oriented model that are usually ignored because of their fast dynamics. The objective of the power management system is to minimize energy consumption while ensuring that all the constraints are enforced.

In this paper, first the fuel economy ceiling of the previously developed 1-state MPC (Model Predictive Control) with only the battery dynamics in the control-oriented model was obtained for a verified high-fidelity model of a Ford power-split HEV. Then by adding the engine speed as an additional state in the MPC control-oriented model, the effects of powertrain inertial dynamics in the MPC power management were modeled and a 2-state MPC is formulated.

The 2-state MPC was optimally calibrated over the high fidelity model of the Ford HEV by using modeFRONTIER. The closed-loop simulation results show that with the proposed 2-stateMPC strategy and without any preview information or any additional engine stop/start strategy, the fuel economy can be improved with respect to the 1-state MPC with a heuristic engine stop/start strategy. This work shows the potential to improve the fuel economy with software changes only. It also presents a systematic approach to the controller parameter calibration.