Otimização da produção de energia elétrica em usinas hidrelétricas

Abstract

Given the current scenario of greater concern regarding sustainable development, renewable energy sources have become increasingly important. Within this context, in addition to contributing to the greatest share of Brazilian renewable generation, hydroelectric power plants are responsible for most of the energy matrix in general. Since different operation strategies can be applied to a power plant’s set of turbines to supply the same power demand, it is presented in this work an investigation of computational intelligence techniques applied to the optimization of the operation of hydroelectric power plants. Such a study is motivated by the potential to save hydro and financial resources, hence contributing to a better operation planning of the electric system, and also by the simplicity in implementation, and effectiveness that intelligent techniques have presented over the years. The following contributions are proposed: (i) an application not yet explored in the literature involving binary computational intelligence techniques, which, basically, benefits from binary algorithms consolidated in the literature to solve problems whose decision variables are integer; (ii) an adaptation of a well known hydroelectric power plants model which allows the maximum saving of hydro resources through a simple modification in the methodology; and (iii) an innovative approach to the fitting of hydraulic turbines efficiency curves, which aims for a more accurate modeling of this component by applying techniques based on the machine learning concept. Simulations were performed with data of a power plant belonging to the EDP group. Adaptations of the Grey Wolf Optimizer and Sine Cosine Algorithm, and the Ant Colony Optimization algorithm have shown to be highly propitious in solving the given problem when duly configured, given the fact that these were able to reliably providing operation schedules that correspond to global optimum.