No label defined (Q4940286)

Let \((\Omega,\sigma,P)\) be a basic probability space, let \(f_{i}\subset\sigma,i=0,1,\ldots,N\), be a family of \(\sigma\)-algebras, let \(H\) be a space of \(f_{i}\)-adapted functions \(x(i)\in R^{n}\), \(\eta(i)\in R^{m}\) be the \(f_{i}\)-adapted random values from \(H\), \(\xi(i)\in R^{m}\) be the \(f_{i+1}\)-adapted random values and let \(\eta(i)\) and \(\xi(i)\) be independent with \(E\xi(i)=0\), \(E\xi(i)\xi'(i)=I.\) The author considers the problem of optimal control of the system \[ x(i+1)=x_0+\sum\limits_{j=0}^{i}a_0(i,j)u(j)+ \sum\limits_{j=0}^{i}a_1(i,j)x(j)+a_2(i)\eta+ \sum\limits_{j=0}^{i}b(i,j)\xi(j). \tag{1} \] An observation process \(y(i)\) is described by the relation \(y(i+1)=A(i)x(i)+B(i)\xi(i)\), \(y(0)=0.\) The functional \({\mathcal J}(u)\) to be minimized is of the form \[ {\mathcal J}(u)=E\Biggl[x'(N)M_0x(N)+ \sum\limits_{j=0}^{N-1}u'(j)N_1(j)u(j)\Biggr]. \] The vector \(x_0\) and the unknown parameter \(\eta\) are mutually independent and independent of the process \(\xi.\) The author finds a relation for the optimal control of the problem (1).