Model Based Predictive Control of Combustion Engine with Constraints

摘要

The paper describes the development of model based predictive control of combustion engine. The basis is the development of nonlinear predictive model that is capable to predict the engine states for several future injection time instances. This model can be identified for a particular real engine. Using this model the future engine behaviour in dependence on possible control inputs (fuel injections, supercharger blade position etc.) are parametrized and using optimisation of performance index the control input is chosen. This approach explicitly enables to control some variables (typically motor revolutions) and simultaneously to fulfill the constraints on other variables (typically lambda).