摘要：AbstractConstraints on control inputs and system outputs are usually given in the non-symmetric form. However, various robust MPC design approaches based on LMIs consider the symmetric representation of the constraints. The aim of the paper is to present a novel concept to overcome this limitation. The approach introduces a time-varying vector of symmetric constraints. Its entries are determined using the switching indicator function. The main advantage is the same computational burden of the solved optimization problem compared to the original procedure. Closed-loop stability is guaranteed either by the feasibility of the modified optimization problem or by a pre-designed support controller when the optimization problem is infeasible. An illustrative example demonstrates that the control performance is significantly improved using the developed strategy.
关键词：KeywordsRobust model predictive controlnon-symmetric constraintslinear matrix inequalitiessemidefinite programmingparametric uncertainty
