For more than a century, both private and commercial vehicles have been powered almost exclusively by internal combustion engines. In recent decades, the emission of pollutants and CO2 have become increasingly urgent issues. To limit pollutant emissions, vehicles are equipped with an exhaust gas aftertreatment system (ATS), while hybridization is used to reduce the fuel consumption and CO2 emissions. As a consequence, modern powertrains are made up of multiple components whose interactions must be carefully managed by a supervisory controller. By exploiting information about the mission, a predictive supervisory controller can optimize the powertrain operation for the current mission, thereby maximizing the powertrain performance. In this thesis, a P3 hybrid electric vehicle (HEV) equipped with a diesel engine featuring a variable engine calibration and an ATS is considered. The resulting degrees of freedom are the gear selection, the motor power, and the engine operation. The task of the supervisory controller is to achieve a charge-sustaining and NOx-compliant operation, while minimizing the fuel consumption over a mission. Four possible powertrain configurations are investigated. Thereby, the goal is the development of a predictive supervisory controller that is transferable to different powertrain configurations.

First, the energy management of an HEV is considered. A fixed engine calibration is assumed and pollutant emissions are neglected, while the motor power and the gear selection are optimized. The benchmark solution shows that the powertrain should be operated either purely electric or purely conventional. A real-time-capable optimization method based on Pontryagin’s minimum principle is developed and employed in a predictive supervisory controller. The controller is validated in simulation, demonstrating near-optimal performance, outperforming a state-of-theart controller, and showing robustness against mispredictions.