Self-bounded controlled invariant subspaces in model following by output feedback: minimal-order solution for nonminimum-phase systems (Q1780596)

Consider the system \(\Sigma_s: x_s(n+1) = A_sx_s(n) + B_s u(n); y_s(t)=C_sx_s(n)\), and the model \(\Sigma_m: x_m(n+1) = A_m x(n) + B_m h(n); y_m(t)=C_mx_m(n)\). It is assumed that the initial states are zero and that both systems are stable. The ``model following by minimal-order dynamic feed-forward'' is to design a stable linear system \(\Sigma_c\) with input \(h\) and output \(u\) such that for all \(h\) the output of \(\Sigma_m\) equals the output of the series connection of \(\Sigma_c\) and \(\Sigma_s\). Furthermore, the state dimension of \(\Sigma_c\) should be minimal. First the authors give a geometric answer to this problem. Hence the answer is written in terms of a largest invariant subspace and of ranges of certain operators. Since for systems it can be hard to calculate these subspaces, the authors next derive more easier checkable conditions under the condition that \(\Sigma_s\) is right-invertible. Furthemore, they obtain useful properties regarding the invariant zeros.