Abstract
In this paper, a search based MPC approach to the cold start control of SI engines is proposed. The combustion of SI engine is a complicated, nonlinear and hybrid process which has four strokes (events) in a cylinder cycle. Each event has different dyanmics. In particular, the control input of spark timing and fuel injection are discrete. For this system, we propose a model predictive control approach based on bisection search for the optimal determination of spark timing. To this end, the highly complicated dynamics is simplied first based on the property of engine combustion, from the view point of torque building. To ensure an effective search, the range of search is also determined based on the property of engine combustion. Application to the SICE engine benchmark shows that this approach works quite satisfactorily.
