Hide menu


Mean-value Observer for a Turbocharged SI-engine

For air/fuel control, where the cylinder air charge (CAC) estimation is based on intake manifold pressure, it is essential to reduce the noise from measured intake manifold pressure signals while at the same time keeping the phase delays low during transients. Observers provide means of filtering without introducing large delays and therefore an observer is developed with emphasis on CAC estimation for air/fuel ratio control. It also enables model based diagnosis opportunities. The observer is based on a nonlinear mean value engine model including the intake and exhaust side together with turbine shaft speed.

The observer feedback gains are determined using an extended Kalman filter. Systematic methods are presented to determine the covariance matrices for the noise and as part of that a novel method is presented to determine the state noise based on the model quality. Additional performance benefits are demonstrated when the model is augmented with one state. The observer is evaluated using real engine data with feedback from measured pressures in the intake manifold and after the intercooler. It shows very good results for CAC estimation as it suppresses intake manifold pressure noise without introducing phase delays. It also provides an accurate estimate of turbocharger speed. Therefore it is highly suitable for control and diagnosis in turbocharged SI-engines.

Per Andersson and Lars Eriksson

IFAC Symposium on Advances in Automotive Control, 2004

Download Article (pdf-file)Show BibTeX entry

Page responsible: webmaster
Last updated: 2021-11-10