On Modeling and Nonlinear Model Reduction in Automotive Systems

Abstract: The current control design development process in automotive industry
and elsewhere involves many expensive experiments and hand-tuning of
control parameters. Model based control design is a promising approach to
reduce costs and development time. In this process low complexity models
are essential and model reduction methods are very useful tools.

This thesis combines the areas of modeling and model reduction with
applications in automotive systems. A model reduction case study is per-
formed on an engine air path. The heuristic method commonly used when
modeling engine dynamics is compared with a more systematic approach
based on the balanced truncation method.

The main contribution of this thesis is a method for model reduction
of nonlinear systems. The procedure is focused on reducing the number
of states using information obtained by linearization around trajectories.
The methodology is closely tied to existing theory on error bounds and
good results are shown in form of examples such as a controller used in
real-world cars.

Also, a model of the exhaust gas oxygen sensor, used for air-fuel ratio
control in automotive spark-ignition engines, is developed and successfully
validated.