A new algorithm for pole assignment of single-input linear systems using state feedback (Q2574669)

The paper presents a numerically reliable algorithm to solve the problem: Given the system \(\dot x=Ax+bu\) and a set of \(n\) complex numbers \(\mathcal{L}\), \(A\in \mathbb R^{n\times n}\), \(b\in \mathbb R^n\). \(x(t)\in \mathbb R^n\) and \(u(t)\in \mathbb R^n\) are the state and the input vectors, resp. Find a vector \(f\in \mathbb R^n\) such that the eigenvalues of \(A+bf^T\) are the poles specified in \(\mathcal{L}\). The algorithm uses the fact that given the vector \(f\), there exists a unitary matrix \(Q\) satisfying \(Q^*(A+bf^T)Q=T\), where \(T\) is upper triangular with eigenvalues of \(A+bf^T\) on its diagonal. \(Q^*\) denotes the conjugate transpose of \(Q\). Multiplying this equation by \(Q\) leads to \(AQ-QT=bx^T\), where \(x=-Q^Tf\). The solution has the form \(f=-\overline Qx\) for given \(x\) and \(Q\), where \(\overline Q\) is the conjugate of \(Q\). The algorithm is based on the Schur decomposition of the closed-loop system and numerically stable unitary or orthogonal transformations which are used whenever possible. Numerical examples show a good general performance in the comparison with MATLAB functions acker and place.