The Bellmann insertions in route problems with constraints and complicated cost functions

Abstract

The route problem with precedence conditions and cost functions depending on the jobs list is considered; these singularities correspond to engineering applications. In particular, the above-mentioned singularities exist in statements of some problems arising in nuclear energetics and in machines with numerical control. Problems involved in sequentially circuiting megalopolises and in carrying out some (interior) work during these circuits are investigated. A procedure for local improvement of heuristic solutions for problems of perceptible dimension is proposed; this procedure exploits insertions on the dynamic programming base. Dynamic programming is realized in the form of a variant that does not provide for construction of a "full" array of values of the Bellman function. The search for localization of an insertion involves restricting to the variant of the Bellman procedure that realizes the extremum of the (local) criterion without constructing a corresponding solution in the form of a route-track pair. A more complete and more cost-intensive (in the sense of memory resources) procedure including determination of the above-mentioned (local optimal) solution is planned after the choice of the insertion localization.