Automating the allocation of water supplies in Texas using an expert system to control the interaction of a geographic information system and an LP solution algorithm.

摘要

A set of procedures to automate regional water resources planning was developed by integrating the capabilities of an expert system, a geographical information system and an LP solver. Water supply and demand data are stored in the database of the geographic information system and used to develop a network of potential allocation arcs linking all supplies to all demands. Planning rules derived from the knowledge and experience of water planning experts are stored in the expert system and used to modify the network. A linear programming solution algorithm is then employed to determine a minimum cost set of flows in the modified feasible network of allocation alternatives.; The system was evaluated on both hypothetical and realistic data sets, and the resulting allocations were demonstrated to be consistent with those produced by existing methods.; Significant reduction in the time necessary to perform an allocation analysis was realized by automating the procedure. In practical applications this can be translated into the ability to more fully explore the effects of variations in supply, demand, transport capacity, and the expert system rules themselves.; The rules, categorized by the scope of their application, were found to be effective in capturing and implementing the considerations of expert water planners.; The most significant advancement to water resources planning afforded by automating the process devolves from the implicit necessity that the process be rigorous. This results in more defensible solutions and in more realistic comparisons of alternative scenarios.