ENGINE REAL TIME MODELS - WAY TO A MODEL BASED PREDICTIVE CONTROL

摘要

The paper deals with a model-based predictive control of a DI diesel engine with Variable Geometry Turbine (VGT). The paper is based on an integrated approach which purpose is to propose a procedure for designing control algorithms. A real-time engine model is the essential part of the work. The model and way how to employ experimental data is described in details in the paper. The proposed real-time model is still yet physically detailed enough. It allows for an extensive testing of control algorithms and designing new ones. The model in combination with the experimental data represents a training data which are used to formulation of a black box model. The black box model is fully real-time capable and it has also the predictive function. The detailed model has been created in the GT-Power code. The simplified engine model has been created in Matlab/Simulink environment. The system identification to obtain the black box model has been performed using the LOLIMOT model. The model application and the model-based engine control are demonstrated on the DI diesel engine for a small truck. The work is mainly focused on the VGT turbocharger control. Results of different control strategies are presented.