The problem of adaptive optimum control in Markov processes

The problem of synthesis of optimum strategy is considered for Dynamic of systems where a part of basic information, required forthe construction of the optimal control is lacking. A method for finding the synthesis of the optimal strategy, based on a recurrent algorithm for computing conditional densitiesis suggested. The basic device at an establishment of the fact of convergence in the given work is the information measure of a mismatch between estimated distribution and an estimation. As such measure so-called information number Kulbaka - Ley-blerais taken. In work convergence of the offered procedure is established.