A novel approach to linear decentralized robust performance stabilization of large-scale systems

The existing methods of decentralized control suffer from two major restrictions. First, almost all of them hinge on Lyapunov's method, and second, they do not address the problem of performance robustness. A novel methodology to overcome the above defects is presented in this paper. Central to this approach is the notion of a finite-spectrum-equivalent descriptor system in the input-output decentralized form. By way of this notion, a new formulation of the interaction which introduces some degrees of freedom into the design procedure is offered. The main result, i.e. a sufficient condition for decentralized performance stabilization in a desirable performance region and maximal robustness to unstructured uncertainties in the controller and plant parameters, nevertheless, is in terms of regular systems. Based on minimal sensitivity design of isolated subsystems via eigenstructure assignment, an analytic method for the satisfaction of the aforementioned sufficient condition is also presented.