scientific article; zbMATH DE number 4005027
Consider the linear controlled object \[ (*)\quad \dot x=A(t)x+B(t)u,\quad u(t)\in \Omega (t), \] where \(x\in R^ n\), A(t) and B(t) are of dimensions \(n\times n\) and \(n\times r\), \(t\in J\), J is an interval \(J\subset (-\infty,\infty)\), \(A(t)\in L^ 1_{loc}\), \(1\leq p\leq \infty\), \(u(t)\in L^ q_{loc}\), \(1\leq q\leq \infty\), \(B(t)\in L^ p_{loc}\), \(1/p+1/q=1\). Here \(A(t)\in L^ 1_{loc}\) means that the elements of A(t) are functions in \(L^ 1_ Q\) for any compact \(Q\subset J.\) For any fixed control \(u(t)\in L^ q_{loc}\), the system (*) has a unique solution \(x(t,x_ 0,t_ 0,u)\), which is an absolutely continuous function satisfying (*) almost everywhere, where \(x_ 0=x(t_ 0)\), \(t_ 0\) is the initial time. The admissible control u(t) is called the reciprocal control on \([t_ 0,t_ 1]\) if only \(x(t_ 1,x_ 0,t_ 0,0)=0.\) The paper deals with necessary and sufficient conditions for controllability to the origin. Both the case of non-bounded and bounded controls is considered. In case of non-bounded control the controllability analysis is rather standard. In the bounded control case controllability of the system (*) can be stated by means of reciprocal controls. Some examples show that in certain practical cases the determination of reciprocal controls can be much simpler than other standard methods.
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3756060)