Optimal experiment design for calibrating an airpath model of a Diesel engine

Mathematical models become more and more indispensable tools for engine manufacturers. As nonlinear dynamic models based on rst-principles are preferred by practitioners, model calibration or parameter estimation is often a time consuming task. The use of optimally designed dynamic inputs can reduce the experimental burden and increase the accuracy of the estimated parameters. The current paper presents the calibration and validation of a Diesel engine airpath model. Optimal inputs have been designed based on random phase multisine inputs. These multisines can be adapted to excite exclusively a speci c frequency band of interest. Moreover, they allow (i) to concentrate the input around an operating point, and (ii) to include fast variations in the input pro le without introducing a large number of discretization parameters. The resulting model has been found to provide an acceptable predictive power in both identi cation and validation.