Design of constrained controls for nonautonomous linear systems (Q1064291)

The linear closed-loop system is given by (*) \(\dot x=A(t)x+u(t,x)\) where \(x: R^ 1\to R^ n\), \(u: R^ 1\times R^ n\to R^ m\), A(t) is of \(n\times n\) size. It is assumed that the feedback function u(t,x): a) is continuous in t and locally Lipschitz in x on the domain \(V\subset E^ 1\times E^ n\); b) for each (t,x)\(\in V\), \(u(t,x)\in U(t)=\{u\in E^ n: u=B(t)w\), \(\| w\| \leq 1\), \(w\in E^ m\}\), dim B\(=n\times n.\) The paper deals with a method of determining a control u(t,x) such that from any state \(x(t_ 0)=x_ 0\in R^ n-\{0\}\) the system (*) reaches the origin in finite time \(T_ 1=T(t_ 0,x_ 0)\). It is proved that if only T(t,x) is locally Lipschitz, then the solution is unique. Moreover, an autonomous closed-loop system of the following type is considered: (**) \(\dot x=Ax+u(x)+g(x,u(x))\) where g(x,u) is locally Lipschitz. It is shown that if in the model (**) the function u(x) is created by the same method as that used for the model (*), the solution of the system (**) exists and reaches the origin in a time \(T_ 1\leq T(x_ 0)\epsilon^{-1}\), where \(T(x_ 0)=T(0,x_ 0)\), \(\epsilon\) is a certain positive constant. The results of the paper have a practical meaning as they can be used for an estimation of the minimal time for a certain class of feedback structures.