Engine Calibration Using Nonlinear Dynamic Modeling

In recent years, engine calibration efforts have increased dramatically in order to fulfill legislation requirements, in particular the reduction of undesired emissions and fuel consumption while maintaining drivability and meeting comfort demands. Dynamic modeling has been shown to be a useful tool in this context. In this paper it is shown how nonlinear dynamic modeling is applied to the virtual calibration of engines. Two different kinds of excitation signals for the training of the models are compared and their usability for engine calibration is analysed. It is shown that virtual engine calibration can lead to a significant reduction of test bench costs during engine and vehicle development while maintaining a high level of calibration accuracy.