Improved Feed-Forward Command Governor strategies for discrete-time time-invariant linear systems

This paper presents a novel enhanced algorithm belonging to the class of so-called Feed-Forward Command Governor (FF-CG) strategies, characterized by the absence of any explicit on-line measure (or estimation) of the state, which have been recently proposed in [1], [2] for input/state constrained discrete-time linear systems subject to bounded disturbances. In fact, while in all traditional CG schemes the set-point manipulation is undertaken on the basis of either the actual measure of the state or its suitable estimation, feed-forward strategies are characterized by the fact that the CG design problem is explicitly solved, with limited performance degradation and with similar properties, in the absence of such a measure. This is achieved by forcing the system evolutions to stay “not too far” from the space of feasible steady-states. Although effective also in the case of bounded disturbances, the performance of earlier FF-CG schemes [1], [2] was mainly limited by the fact that such strategies were instructed to maintain constant their actions for a prescribed number of sampling steps between two subsequent CG action computation. Here such a restriction is completely removed and a novel class of FF-CG schemes able to update their actions at each sampling step is discussed. Finally, in order to evaluate the effectiveness of the proposed solution, numerical simulations on a physical plant have been undertaken and comparison results reported.