Bayesian Inferencing Applied to Real‐Time Reservoir Operations

Abstract

A significant amount of research during the past few years has focused on the application of expert systems technology to problems of water resources management. While these investigations have led to speculation as to the benefits of intelligent reasoning applied to real‐time reservoir operation, working systems are nonexistent or in the preliminary stages of development and testing. This research presents a novel perspective on the use of knowledge‐based inferencing techniques applied to real‐time reservoir operation. A hybrid reasoning structure using both Bayesian and rules‐based inferencing is presented. Rules are used to achieve a real‐time simulation that is comparable to other rule‐based systems reflecting expert operations as proposed in the literature. The Bayesian mechanism then provides a judgment about the quality of recommended releases based on prior information and present conditions. An additional feature of this system is its learning capabilities that can be used for further refinement of system recommendations.