Algorithm for an optimal observation problem with non-convex a priori state distribution (Q1342570)

Optimal observation problems in control theory, which are no longer linear, are considered. In such problems the unknown initial state of a dynamic system may belong to one of a number of given convex sets. Incomplete and inaccurate measurements by a static observer are assumed. The problem is to identify both the set which contains the state, having generated the observed process, and the mentioned state itself. The procedure is a generalization of the technique of \textit{R. Gabasov} and \textit{F. M. Kirillova} [Dokl. Akad. Nauk BSSR 34, No. 9, 777-780 (1990; Zbl 0716.93004)]. An optimality criterion is discussed. An algorithm is proposed to solve the problem. It is based on sequential reduction of the domains where the main constraints of the problem are not satisfied. A numerical experiment is carried out, the results are tabulated and plotted.