On the effect of prediction model complexity on the performance of a nonlinear model predictive controller for the energy management of a parallel hybrid electric vehicle

This master’s thesis work deals with the control system development and performance evaluation for a through-the-road-parallel (TTRP) hybrid electric vehicle with an objective of reducing the fuel consumption and emission. A conventional 2.0 l diesel ICE vehicle is downsized to 1.6 l diesel engine and an electric motor is integrated at the rear axle allowing the implementation of through-the-road-parallel hybrid architecture. Further, a motor generator unit is coupled with the ICE at the front axle to implement a start and stop control strategy. The aim of this work is to study the effect of prediction model complexity on the performance of a nonlinear model predictive controller for the energy management of a downsized through-the-road-parallel hybrid electric vehicle, demonstrating the viability in terms of fuel economy and emission performance. An adaptive equivalent consumption minimization strategy (A-ECMS) and a nonlinear model predictive control (NMPC) with different levels of the prediction model complexity are proposed for the hybrid vehicle along with the imposition of optimal gear shift selection. The A-ECMS and NMPC are validated numerically for different driving cycles against the conventional 2.0 l diesel ICE, maintaining the battery state-of charge within permissible level. The tabulated results show that the through-the-roadparallel hybrid vehicle performs better in terms of fuel consumption and CO2 emission against the baseline vehicle. The NMPC performs better in terms of fuel saving when compared to the A-ECMS and could achieve a greater fuel saving adopting a more complex prediction model which reflects more accurately the real vehicle behaviour. In addition, an analysis and minimization of pollutants and emissions was carried out, in the worst case scenario without the use of the electrically heated catalyst (eHC), along two driving cycles: WLTP and RDE (supplied by IDIADA). The final RDE emissions results, calculated with the method of the moving averaging window, show that the limit of NOx is fully respected with the optimal torque split through the implementation of the two controllers ECMS and NMPC, and the use of the exhaust model in GT-Suite for the after-treatment.