Approximation of Feedback Gains Using Spectral Projections: Application to the Oseen System
Mehdi Badra,Jean-Pierre Raymond
DOI: https://doi.org/10.1137/23m1609695
IF: 2.2
2024-11-06
SIAM Journal on Control and Optimization
Abstract:SIAM Journal on Control and Optimization, Volume 62, Issue 6, Page 2910-2935, December 2024. We consider a parabolic controlled system (called [math]) in a Hilbert space [math] and a family of approximate parabolic controlled systems (called [math]) in a Hilbert space [math]. We are in the case of nonconforming approximations, that is, when [math]. We assume that both [math] and [math] are Hilbert subspaces of another Hilbert space [math]. We assume that [math] and [math] satisfy approximation errors, [math] is exponentially stabilizable, and [math] is uniformly parabolic. We define a reduced order model, called [math] (resp., [math]), based on spectral projections for [math] (resp., [math]). We construct a Riccati based feedback [math] for [math] which exponentially stabilizes both [math] and [math], and we construct an associated Riccati based feedback [math] for [math] which exponentially stabilizes both [math] and [math]. We prove convergence rates for [math], and we prove that, for all [math], where [math] is small enough, [math] also exponentially stabilizes [math]. We apply these results to the boundary feedback stabilization of the Oseen system (the linearized Navier–Stokes system around an unstable stationary solution) and to its numerical approximation by a finite element method (F.E.M.). In that case [math], where [math] is the meshsize of the F.E.M.
mathematics, applied,automation & control systems