Reservoir Operation Using Bayesian Inferencing and Balancing Rules

The water resources problem addressed in this paper is real-time, multi-purpose reservoir operation. Specifically, daily, or at the most, weekly operations are considered. A hybrid modeling scheme is proposed. It entails the use a probabilistic dynamic programming scheme (which could include updating) and a Bayesian inferencing component. The model could be used as a simulation tool as well as a decision support module. Although intelligent reasoning models have been proposed in the literature, this one differs by the fact that the optimization component is closely coupled to the reasoning sub-model. Hydropower production as well as flow augmentation, flood flow attenuation and water supply are considered in the model development.