Decentralized Robust Control of Linear Uncertain Systems

This paper is concerned with decentralized control design for uncertain linear systems; the main aim is to reduce computational complexity and/or problem size. Two approaches to robust decentralized control are studied and compared. First approach is based on the structured (decentralized) controller design for the whole system model. The novel robust stability condition in the form appropriate for convexification is developed and the respective LMI based design procedure is proposed. Second approach is based on subsystems robust controllers, where the interaction bounds are considered in local controller design. The decentralized control problem then reduces to a solution of robust control problems of subsystems'size. Both approaches has been tested on examples, one of them is presented in the paper.