This content is not included in your SAE MOBILUS subscription, or you are not logged in.
NARMAX Structure Selection for Powertrain Control
Published April 19, 2004 by University of Salerno in Italy
Powertrain control models must often be acquired as nonlinear black- box models. This paper discusses a method developed for structure detection and parameter estimation, for models belonging to the AR family and which are estimated using the Gaussian Least Squares algorithm. The process includes both parametric and non-parametric identification tests on the input-output relationships of the estimation data. The suitability for inclusion in the model of all terms within prescribed maximum orders is assessed through statistical calculations and end model performance procedures. The approach is applied to an SI engine MISO identification.