Metaheurísticas para problemas de otimização em dois níveis

• Autor(a) principal: Angelo, Jaqueline da Silva
• Data de Publicação: 2014
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações do LNCC
• Texto Completo: https://tede.lncc.br/handle/tede/209
• topic: Otimização matemática; Otimização em dois níveis; Metaheurística; Mathematical optimization; Metaheuristics; CNPQ::CIENCIAS EXATAS E DA TERRA::MATEMATICA::MATEMATICA APLICADA::ANALISE NUMERICA

This work aims at the development and implementation of robust and efficient computational algorithms to treat multilevel optimization problems, particularly bilevel problems. Those problems are characterized by an optimization problem within the constraints of another optimization problem, and are considered more difficult to treat than classical optimization problems, since, in general, they are non-convex nor differentiable, even when the functions involved are all linear. To solve those problems, different techniques were developed which are based on Ant Colony Optimization and Differential Evolution metaheuristics. Beside those, a surrogate model (metamodel) was also developed, based on the Nearest Neighbors Method, in an attempt to reduce the computational cost of one of the proposed methods. A variety of bilevel problems were addressed to validate the proposed algorithms, including: (i) optimization problems in continuous space with and without constraints; (ii) an application in Operational Research involving the production and distribution planning problem; and (iii) bilevel problems containing multiple followers in the lower level. The analysis of the applicability and the performance of the proposed methodologies showed that they were able to successfully solve all problems, in which competitive results were obtained concerning the applications addressed.