Title:Nonlinear Model Predictive Control for a Diesel Engine AirpathAuthor(s):Roland BergmannAbstract:In today's society, passenger cars play an important role as most widespread means of transportation. The number of cars driving along the roads increases every year. In Austria, especially the number of vehicles powered by diesel engines rises continuously. But more cars also cause more air pollution. So the government restricts the permitted emissions in periodically intervals to lower levels. This is the main reason for car manufacturer to work persistently on the improvement of modern engines. The aim is to reduce emissions while the demand for available power increases. Several inventions like exhaust gas recirculation (EGR) and variable geometry turbochargers (VGT) made this progress possible. Further restrictions on emissions will call for improved engine control structures too. The aim of this diploma thesis is to design a Nonlinear Model Predictive Controller (NMPC) for the airpath control of a diesel engine equipped with two actuators, called exhaust gas recirculation (EGR) valve and variable geometry turbocharger (VGT). Therefore, a suitable model structure for an output prediction model, which is required by the NMPC, must be investigated. Hence, two different modelling techniques are studied. A model structure mainly based on physical equations, named Mean Value Model (MVM) on the one side, and a data based technique for the creation of a so called Linear Parameter Varying (LPV) model on the other side. The required computational power for the online calculation of the manipulated variables is the limiting factor in this engine control application. This nonlinear approach is intended to characterise the airpath behaviour in the engines whole operation area by means of one plant/controller combination only, in distinction to a previous work where several linear models/controller combinations were used. This work already showed that a further reduction of engine emissions is possible by applying alternative, more efficient control structures.Publishing:5/2008