Residence probability control (Q914632)

Consider the linear control system \[ (1)\quad dx=(Ax+Bu)dt+\epsilon Cdw,\quad x(0)=x_ 0, \] where x is the state n-vector, u is a control m-vector, and w(t) is a standard r-dimensional Brownian motion and \(0<\epsilon \ll 1\). The problem is to maintain \(x(t,x_ 0,u)\), the solution of equation (1), in a given set \(\Omega\) during a specified time interval [0,T]. Let \(\tau^{\epsilon}(x_ 0,u)\) denote the first passage time in \(\Omega\), where \(x_ 0\in \Omega\), and consider the condition \[ (2)\quad P_{x_ 0}(\tau^{\epsilon}(x_ 0,u)\geq T)\geq 1-\delta,\quad 0<\delta <1. \] There is shown that linear systems of form (1) with linear state feedback control laws can be divided into two classes, weakly and strongly residence probability controllable systems. Here, weakly residence probability controllable are these systems for which condition (2) can be satisfied for some \(0<\delta <1\) and strongly residence probability controllable are systems for which (2) holds for any \(0<\delta <1\).