Robust pole assignment techniques via state feedback

Abstract

We present a unifying computational framework to solve robust pole assignment problems for linear systems using state feedback. The new framework uses Sylvester equation based parametrizations of the pole assignment problems. The non-uniqueness of solutions is exploited by minimizing additionally sensitivity of closed-loop eigenvalues and the norm of the corresponding state feedback matrix. The solution methods rely on using gradient search based minimization techniques on suitably defined cost functions. The discussion of main functional and numerical aspects reveals many desirable features of the underlying algorithms which recommend them to serve as bases for robust numerical software implementations.