Stability of infinite-dimensional sampled-data systems (Q4806486)

The paper is concerned with the following system with state space \(X\) and input space \(U\) (both Hilbert spaces): \[ \dot{x}(t)=Ax(t)+Bu(t),\qquad x(0)=x^{0}\in X,\tag{s1} \] where \(A\) is the generator of a strongly continuous semigroup \(T(t)\) on \(X\) and \(B\) is a bounded linear operator from \(U\) into \(X_{-1},\) where \(X_{-1}\) is a certain extrapolation space with \(X\hookrightarrow X_{-1}.\) The fundamental problem is stated as follows. Assume that the feedback control \(u(t)=Fx(t),\) where \(F\) is a bounded linear operator from \(X\) into \(U,\) is an exponentially stabilizing state feedback control for (s1) in the sense that \(A+BF\) generates an exponentially stable strongly continuous semigroup on \(X.\) Will this sampled-data controller stabilize the system for all sufficiently small sampling times? The answer is affirmative if \(X\) and \(U\) are finite-dimensional spaces. The authors show that if the feedback operator \(F\) is compact and the semigroup generated by \(A+BF\) is exponentially stable, then, for sufficiently small \(\tau>0,\) the sampled-data feedback \[ u(t)=Fx(k\tau),\qquad t\in[k\tau,(k+1)\tau)\tag{s2} \] is stabilizing. A simple example which shows that the result fails to hold if the compactness assumption on \(F\) is removed is given. Then the case where the input operator \(B\) is unbounded and the semigroup generated by \(A\) is analytic is considered. The authors show that for any compact feedback operator \(F,\) the operator \(A+BF\) (with suitable domain) generates an analytic semigroup on \(X,\) and if this semigroup is exponentially stable, then the sampled-data feedback (s2) is stabilizing for all sufficiently small \(\tau>0.\) Finally, it is demonstrated how the static state feedback results mentioned above can be used in the practically more relevant situation of sampling a dynamic output feedback controller.