Model-based predictive control of a three-way catalytic converter

A high performance of a three-way catalytic converter during engine transients can be achieved by controlling the level of oxygen stored on the ceria surface. The system is controlled by using the process model as the inferential sensor. Desired control actions are calculated off-line by the model predictive controller. These actions are then used to train an analytic neuro-fuzzy controller. The controller can thus cope with the necessary speed for the online calculation yet preserving the high performance of the model predictive controller.