USE OF MONTE CARLO OPTIMIZATION AND ARTIFICIAL NEURAL NETWORKS FOR DERIVING RESERVOIR OPERATING RULES

This paper deals with the use of Monte Carlo optimization and Artificial Neural Networks (ANNs) for deriving monthly reservoir operating rules. The procedure generates synthetic inflow scenarios which are used by a deterministic optimization model to find optimal releases. The ensemble of optimal release data is related to storage and inflow in order to form allocation rules. Different from the common use of regression analysis to define equations relating releases to the other variables, this paper uses ANNs to calculate the releases to be implemented at each period. Simulations based on the use of numerical interpolation instead of ANNs are used for comparison. The procedure is applied to the multipurpose reservoir that supplies the city of Matsuyama in Japan and the results show high correlation with those using optimization under perfect forecast.