A computational algorithm for pole assignment of linear multiinput systems

Abstract

An efficient computational algorithm for pole assignment of linear multiinput systems is proposed. A preliminary stage of the algorithm is a reduction of the system matrices into orthogonal canonical form. The gain matrix elements are then found by orthogonal transformation of the closed-loop system matrix into upper quasi-triangular form whose diagonal blocks correspond to the desired poles. The algorithm is numerically stable, the computed gain matrix being exact for a system with slightly perturbed matrices. It works equally well with real and complex, distinct, and multiple poles and is applicable to ill-conditioned and high-order problems.