Generic Pole Assignment via Dynamic Feedback (Q4388918)

Let \(G(s)\) be a strictly proper transfer function of degree \(n\), with \(m\) inputs and \(p\) outputs. We say that \(G(s)\) is pole assignable by degree \(d\) compensators if for any polynomial \(P(s)\) of degree \(n + d\), one can choose a proper transfer function \(K(s)\) of degree at most \(d\) such that the characteristic polynomial \(\chi '_{cl}(s)\) of \(G_{cl}(s) = (I_{p} - G(s)K(s))^{-1}G(s)\) coincides with the polynomial \(P(s)\). The generic pole placement problem is to find the minimum \(d\) for almost all \(G(s)\). Sufficient conditions for generic pole placement have been studied and the best known result, proved by \textit{X. Wang} and \textit{J. Rosenthal} using algebraic geometry [31st IEEE Conference on Decision and Control, 3098-3099 (1992)], states that \(n < mp + d\max (m,p).\) In the present paper it is shown that the same result can be obtained in a much simpler way, without using algebraic geometry. For generic pole placement, the necessary condition is that \(n \leq mp + d(m + p - 1)\), proved by \textit{J. Rosenthal} [SIAM J. Control Optimization 32, No. 1, 279-296 (1994; Zbl 0797.93018)] also using algebraic geometry. Again, the author shows that algebraic geometry is not needed and that a group action may be used to prove the result. This introduces a new approach which may have some potential for future investigations.