No label defined (Q4533133)

The author considers a stochastic large scale linear time-invariant discrete-time system of the form \[ \begin{aligned} X_{i}(n+1)&=A_{i1}X_{1}(n)+A_{i2}X_{2}(n)+B_{i}U(n)+G_{i}W(n),\\ X_{i}(0)&=X_{i0},\\ Y(n)&=C_{1}X_{1}(n)+C_{2}X_{2}(n)+V(n),\end{aligned} \] where \(i=1,2, X_{1}\) are the slow states, \(X_{2}\) are the fast states, \(U\) is a deterministic input, \(Y\) is a vector of observations, \(W\) and \(V\) are independent Gaussian processes. The goal is to find the control \(u(n)\) that minimizes the performance index. The behavior of the optimal control and the performance index are investigated. It is demonstrated that the optimal control can be approximated by a near optimal control, which is obtained as a combination of a slow and fast control. Finally, a suitable algorithm for computer simulations is presented. For the entire collection see [Zbl 0982.00035].