Projeto multicritério de sistemas de distribuição de energia elétrica sob contextos incertos utilizando algoritmos de busca local

Abstract

The continuous use of optimization techniques for electrical energy distribution project is justified by its social and economical importance since it attends a great amount of Brazilian people. Some algorithms can not be used to obtain the solution for the energy distribution network project due to the complexity of the problem. Evolutionary Algorithmsare an alternative to solve the problem due to their robustness, flexibility and capacity of adaptation for several kinds of problems. The NSGA-II Algorithm is used here to obtain efficient solutions for the problem of electrical energy distribution network considering the most likely load scenario taking into account three objectives (cost, reliability and robustness). The uncertainties in the project are treated through the evaluation of those solutions for different scenarios created by Monte Carlo Simulation. The data obtained using the Monte Carlo Simulations are used as an input for a Multiobjective Sensitivity Analysis where the solutions are evaluated for six distinct criterions. A set of non-dominated solutions is then obtained and a local search method developed in this work is applied for those solutions. The local search method is based on the generation of random networks at pre defined distances using the T-norm metric. This enables the generation of a bigger number of robust solutions for the electrical energy distribution network problem. It is provided to the network designer a large range of possibilities for choosing the better network to be installed. Important aspects for operating the network have been taken into account such as reliabilityand energy losses. A robustness standard is previously incorporated to the network set of possibilities given to the designer