Extension of a Linear Optimal Control Strategy for HEV

Since the last decade, hybrid electric vehicles have been and are introduced increasingly by automotive industry as they provide substantial improvements in fuel consumption. However, the optimal power distribution between the different energy sources in a HEV to achieve high efficiency is a non trivial problem. In a former work, an approximate solution to the generic optimization task by a two-stage approach based on linear programming and switching strategies was introduced, which is extended within this work. First, the opportunity to declutch and switch off the internal combustion engine is considered, which extends the linear program to a mixed integer linear program and leads to additional improvements in fuel consumption. The second extension introduces a receding horizon strategy that considers the current and a target state of charge of the battery at the end of the horizon during the top level optimization. Thus a complete knowledge of the driving cycle is not required but only a prediction for a shorter horizon. Furthermore, errors caused by model-plant mismatch and prediction can be compensated by the SOC feedback. Both extensions result in a minimum increase in computational effort and allow to formulate the optimization as mixed integer linear program and solve it by real-time capable solvers. As application example a parallel hybrid electric vehicle with realistic models of battery and powertrain and the opportunity to declutch the internal combustion engine is considered. The method and the proposed extensions are evaluated in simulation and experimental in which satisfactory results could be achieved in both cases.