Abstract
This paper proposes a design method of robust servosystems. The plant we consider is assumed to have structured uncertainties which are represented by matrix polytopes in the state space equation. We construct the control system as follows. First, we quadratically stabilize the control system containing a integral compensator and guarantee to track a constant reference signal. Next, closed-loop poles are clustered robustly in a specified region. Further, we give an upper bound of tracking errors of controlled output and control input in L2 norm sense and minimize it subject to the uncertainties. To construct a compensator, we parametrize stabilizing compensators and reduce the above design problem to a convex optimization on the parameter space.
