| Popis: |
Improvements in the availability of sensed and actuated vehicular systems have increased the opportunity for applying automatic control. Classic methods for designing control architectures are no longer sufficient due to the highly non-linear behaviour and heavy coupling between sub-systems. As a result, there is scope to employ computationally powerful optimisation techniques to investigate design trade-offs in these increasingly complex systems. This thesis intends to take a multi-faceted look into the use of powerful optimal control techniques to solve some of the control problems arising in the automotive sector. In particular, a number of different control problems focused on emissions management are considered. The first scenario relates to the fuel optimal control of a road vehicle, with a specific focus generating fuel-optimal drive cycle trajectories incorporating emissions constraints. A simplified model will be used initially to examine the solution structure and highlight any important features which occur in the solving of such eco-driving problems. Additionally, there are comparisons to be made between using a continuously variable transmission and a more traditional discrete gearbox. The second scenario considers an optimal driving problem for a vehicle equipped hybrid electric powertrain with a thermal aftertreatment model. A large percentage of tailpipe emissions are produced when the three-way catalytic converter is cold. Increased flexibility of power delivery enables the engine to be powered greater than the tractive demand, which prompts quicker heating and thus producing less total emissions. The drawback to this method is an increase in the tailpipe flowrates during this heating period. The possibility and scope for the reduction of emissions is considered. The final scenario considered in this thesis will cover the optimal control of gasoline engines with combustion mode switching events. In particular, a turbocharged gasoline engine equipped dual-independent variable valve timing is considered. For engines operating in a stoichiometric regime, three-way catalytic converters are effective at reducing emissions. If the combustion mode is switched to lean, there is an increase in the tailpipe nitrous oxides (NOx). In order to avoid this, a lean-NOx trap is installed in the tailpipe, which requires periodic regeneration to ensure its effectiveness. Dynamic compensation of all actuators vastly improves the performance of the switching control. This thesis presents a number of suitable modelling techniques to investigate these scenarios, the results of which can be used to aid in the design of realtime control algorithms. |
