Approximation of norm-optimal controllers

by means of an application-friendly network of widely used control modules

For systems with several inputs and outputs we often have the problem that under certain operation conditions the controller for the different system variables act against each other. To avoid this problem, one could use a multi-dimensional state-space controller, but this type of controller is so complex, that in case of changes in the system it cannot be tuned any more. Therefore, at the IAT of Bremen University an algorithm is currently developed, that provides an optimal PI(D) controller structure including parameterization for such systems. The optimization means a maximization of control quality by minimization of the -Norm. From that point of view, the algorithm can be seen as a link between modern control theory and practical application.