Operation Planning of Hydroelectric Systems: Application of Genetic Algorithms and Differential Evolution

Abstract

The Operation Planning of Hydroelectric Systems is a large, dynamic, stochastic, interconnected and nonlinear optimization problem. In this model, the minimization of penalized thermal complementation is considered as the objective function with the water discharge of hydroelectric plants at each period as the decision variables. An adaptation of two Evolutionary Metaheuristics, the Genetic Algorithm and the Differential Evolution, are proposed in this paper to solve this problem. These methods consider a set of solutions in order to perform exploration and exploitation of the search space allowing them to find several good quality solutions that can serve as alternatives to a given scenario. Tests performed with the Brazilian Subsystems and compared to one of the current used approaches show that the evolutionary methods can improve current solutions and can also bring the benefit of alternative solutions.