Optimal technique for estimating the reachable set of a controlled n-dimensional linear system

Abstract

The reachable set from a given point of a controlled dynamical system is the set of all states to which the system can be driven from that point in a finite time by the allowed controls. A technique is presented Tor estimating the reachable set from the asymptotically-stable origin of a class of n-dimensional linear systems under bounded control. The technique is an optimal version of a Lyapunov method, and provides an (over-)estimate of the full reachable set; it involves the minimization of a quadratic constraint, followed by the maximization of a quadratic form subject to this constraint. The non-linear optimization problems can be routinely solved by means of computer algebra and commonly available computer software. In general, the technique produces a much-improved estimate of the reachable set compared to that given by the standard Lyapunov method. Another advantage of the technique is that it is truly applicable to higher-dimensional systems (n ≥ 3). Since the estimate produced is in the convenient form of an n-dimensional ellipsoid, projections of the estimate onto any space of dimension ≤ n — 1 can be readily found. Problems of two, three, and four dimensions are solved to illustrate the technique.