PIPING NETWORK DESIGN UNDER THE CONDITION OF SUBDEFINITE AND INCOMPLETE DATA

摘要

A new approach is suggested for the solution of direct and inverse problems in hydraulic engineering with imperfect data under presence of set of constraints of various kinds. This approach is based on artificial intelligence techniques. The paper summarizes many years' results on adaptation of artificial intelligence technique for hydraulic design. The applications of such techniques as the subdefinite computational models, the constraint programming, etc, are considered in the paper. Their advantage consists in the following: they a) allow use the subdefinite and incomplete data; b) allow to integrate the heterogeneous constraints by a natural way; c) allow equally simply to consider both direct and inverse problems; d) require description of computational model only, what eliminates the need for algorithm design. The using of aforesaid AI-techniques is illustrated by examples of solution the design problems of pipeline and piping network. The problem of piping networks' optimal designing is among them. The classification of the problems is suggested, which considers the direct well-conditioned and ill-conditioned problems and inverse problems. The results of numerous computational experiments are discussed.