Robust controller synthesis via non-linear matrix inequalities

From MaRDI portal





The authors propose a continuation algorithm for the solution to nonlinear matrix inequalities arising in robust control systems synthesis based on mixed structured singular value analysis. More precisely the proposed method of robust controllers design is based on the stable factors of the multipliers and does not require their restriction to a basis as is the case in the bilinear matrix inequalities approach [\textit{M. G. Safonov}, \textit{K. C. Goh}, and \textit{J. H. Ly}, ``Control systems synthesis via bilinear matrix inequalities, Proc. ACC 1994, 45-59]. The algorithm converts the matrix inequality problem to an ordinary differential equation problem solved using correction/prediction techniques followed by a number of correction steps, which require a series of linear matrix inequalities to be solved. The algorithm may in general converge to a local minimum, but the authors claim that initializing it based on physical insight reduces such danger. The approach is illustrated by a two-mass/spring benchmark example, which requires the solution to a total of 121 linear matrix inequalities.











This page was built for publication: Robust controller synthesis via non-linear matrix inequalities

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4485277)