Regras operacionais para reservatórios baseadas em otimização estocástica implícita e redes neurais de função de base radial.

Abstract

In this work, we present monthly operating rules based on Implicit Stochastic Optimization (OEI) and Radial Basis Function Neural Networks (RBF) for Coremas – Mãe D’Água reservoirs, which are located in a semiarid land of Paraiba State, Brazil. The OEI technique consists of optimizing the operation of the system considering a set of possible inflow scenarios, and then using the optimal data for constructing reservoir operating rules. The synthetic inflow scenarios were obtained by the Method of Fragments. In this study, the RBF was used to relate reservoir initial storage and current in flow with allocation coefficients for each month of the year. The monthly operating rules obtained with the OEI-RBF model were applied to the operation of the water system and a criterion of vulnerability was used in order to analyze the results. According to the vulnerability results, it can be concluded that the EIO-RBF model was superior to standard rules of operation.