Interplay between discretization and controllability of linear delay systems: an algebraic viewpoint

Linear functional-differential equations (34K06) Control problems for functional-differential equations (34K35) Discrete version of topics in analysis (39A12) Controllability (93B05)

Review text

The authors investigate controllability of delay-differential systems and its preservation through time discretization. The delay systems under study are linear time-invariant with a punctual delay \(\tau>0\), governed by\N\[\NE(\delta_{\tau}) \dot{x} = A (\delta_{\tau}) x + \sum_{j=1}^k B_j(\delta_{\tau}) u^{(j)}, \N\]\Nwhere \(x \in \mathbb{R}^n\), \(u\in \mathbb{R}^m\), \(E\), \(A\), \(B_1,\dots,B_k\) are matrices of appropriate sizes with real polynomial coefficients, \(E\) is of full rank and \(\delta_{\tau}\) denotes the delay operator associated to the delay \(\tau\).\N\NThe analysis is carried out in a module-theoretic framework which side-steps existence and uniqueness issues to focus in structural properties. Several concepts of controllability, are expressed in this context which are equivalent in the finite-dimensional case but not in the presence of delays. This first part is essentially self-contained and can provide a useful introduction to this algebraic approach.\N\NThe second part is dedicated to the study of discretizers. Given a discretization time \(h > 0\), which is required to divide \(\tau\), the discretizers under study correspond in the Laplace/\(z\) domain to the replacement of \(s\) with \(p(z^{-1})/(hq(z^{-1}))\) where \(p\) and \(q\) are polynomials (for example: \(s\) replaced with \((1-z^{-1})/(hz^{-1})\) for the explicit Euler scheme). An asymptotic condition on the discretization is added to ensure that it is asymptotically faithful for small values of \(h\). The authors formalize this class of discretizers as functors between the categories of delay-differential systems with delay \(\tau\) and the category of LTI discrete-time system of delay \(h\).\N\NFinally, the authors prove that for a given discretizer, there is a delay system for which even the weakest controllability property is not preserved but that conversely, for any given delay system, there is an appropriate discretizer which preserves controllability.

Cites work

This page was built for publication: Interplay between discretization and controllability of linear delay systems: an algebraic viewpoint

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7005494)