A system inversion approach to robust set-point regulation

We present a novel methodology, based on system inversion, for the set-point regulation of a system with structured uncertainties. The approach consists in choosing a polynomial as the output function and then designing both the controller and the reference input in order to minimize the worst-case settling-time of the set-point regulation, with constraints on the absolute value of the control variable and on the maximum overshoot of the output. A comparison with a traditional methodology, based on the step-response, shows how the overall design is simplified and how the worst-case settling-time can be greatly reduced. Optimization has been performed by means of genetic algorithms.