A novel nonlinear model reduction method applied to automotive controller software

The automotive industry is experiencing tightening emission legislations together with high demands on performance and driveability. As a counteraction, controller software tends to become more and more complex. Intricate controller software has several downsides, the large number of controller parameters yields an exhaustive calibration task, often performed through costly experiments. In addition, to guarantee reliability, validation and verification analysis is performed on the controller in combination with the engine. This task would also greatly benefit from a less complex controller structure. Here a novel model reduction method of nonlinear discrete-time systems is introduced and applied to an engine controller used in current production cars. The result is a nonlinear piecewise affine system with improved simulation speed.