Finite-dimensional control of distributed parameter systems by Galerkin approximation of infinite dimensional controllers (Q1078132)

The author considers finite-dimensional control of the linear distributed parameter system \(\partial v(t)/\partial t=Av(t)Bf(t)\), \(v(0)=v_ 0\), \(y(t)=cV(t)\), where the state v(t) is in an infinite-dimensional Hilbert space, A is a closed linear unbounded operator with domain dense in H and the input-output operators B and C have finite ranks M and P respectively, f(t) and y(t) representing the inputs for M linear activators and the outputs from P linear sensors, respectively. Two basic approaches for model reduction of this system via the Galerkin method are given. A direct approach approximates the open-loop system and then generates the controller for this approximation; an indirect approach approximates the infinite-dimensional stabilizing controller. Conditions are given under which the two approaches are equivalent in the sense that they both produce finite-dimensional controllers which stabilize the infinite-dimensional system. The indirect approach is shown to converge to the stable closed-loop system consisting of the distributed parameter system and an infinite-dimensional controller. Conditions are presented on the behaviour of the Galerkin method for the open-loop system which guarantee closed-loop stability for large enough finite-dimensional approximations.