COMPUTER-AIDED DESIGN OF SUBOPTIMAL CONTROLLERS FOR LARGE-SCALE SYSTEMS.

A procedure is described for designing a decentralized control system for a large-scale dynamical system. Constraints are imposed on the controller structure as well as on the information set of each controller. The emphasis here is on the computational aspects of the design problem. The computational technique for finding the optimal feedback matrices appears to be especially effective when the system is composed of several weakly coupled subsystems. Also the results given here may suggest a more efficient method for solving the single-controller problem by decomposing it into smaller coupled problems.