Title :
Controller design for mixed l1/LMI performance objectives
Author :
Young, Peter M. ; Dahleh, Munther A.
Author_Institution :
Lab. for Inf. & Decision Syst., MIT, Cambridge, MA, USA
Abstract :
In this paper we study norm minimization problems, and in particular the l1 optimal control problem, subject to linear matrix inequality (LMI) constraints. We show how LMI´s can be recast as parametrized LP´s, and from this develop a sufficient condition for the lack of a duality gap for such problems in the infinite dimensional case. We give an exact characterization of the primal and dual problems in this case. In the case that we use the l1 norm then both primal and dual problems can be posed as LMI optimization problems
Keywords :
control system synthesis; duality (mathematics); linear programming; matrix algebra; optimal control; LMI optimization problems; controller design; l1 optimal control problem; linear matrix inequality constraints; linear programming; mixed l1/LMI performance objectives; norm minimization problems; parametrized LP; Constraint optimization; Design methodology; Ear; Kernel; Linear matrix inequalities; Linear programming; MIMO; Optimal control; Symmetric matrices; Vectors;
Conference_Titel :
American Control Conference, Proceedings of the 1995
Conference_Location :
Seattle, WA
Print_ISBN :
0-7803-2445-5
DOI :
10.1109/ACC.1995.520987
