A Plug-in Hybrid Electric Vehicle (PHEV) can achieve a considerably higher overall fuel economy than conventional vehicles. The fuel economy of PHEVs however, strongly depends on the supervisory control strategy of the hybrid powertrain. Compared to conventional, non-predictive, supervisory control systems, predictive control methods can further improve the fuel economy by anticipating on future driving conditions. This thesis presents a predictive optimal control method for PHEVs that minimizes the fuel consumption in real-time operation, based on the preview of the future topographic profile and velocity profile. This strategy is realized by a global estimation method that determines the optimal battery depletion strategy on the global range, which is tracked by a novel adaptive heuristic control method that optimizes the driving mode and power distribution between the two power sources on the local range. Compared to state of the art predictive control methods, this control method makes it possible to include feasibility and drivability conditions (such as the number of engine starts and variation of the torque demand) in the control strategy, while allowing high update frequencies of the control actions.

The proposed control method is fine-tuned on a P2 parallel PHEV model based on a Volkswagen Jetta HEV, and its effectiveness is evaluated by implementation in a high fidelity simulation environment. It was found that route-preview based control strategies can obtain substantial improvements of up to 7.3% in fuel economy compared to a non-predictive heuristic control method. Simulation results also demonstrate that the application of optimization based control methods without any drivability considerations (such as A-ECMS) often leads to unacceptable control policies. The proposed control method on the other hand, was able to take feasibility limitations into account, which led to a reduction of the number of combustion engine starts of up to 52%.