Publicação

On Challenging Techniques for Constrained Global Optimization

Detalhes bibliográficos
Resumo:	This chapter aims to address the challenging and demanding issue of solving a continuous nonlinear constrained global optimization problem. We propose four stochastic methods that rely on a population of points to diversify the search for a global solution: genetic algorithm, differential evolution, artificial fish swarm algorithm and electromagnetism-like mechanism. The performance of different variants of these algorithms is analyzed using a benchmark set of problems. Three different strategies to handle the equality and inequality constraints of the problem are addressed. An augmented Lagrangian-based technique, the tournament selection based on feasibility and dominance rules, and a strategy based on ranking objective and constraint violation are presented and tested. Numerical experiments are reported showing the effectiveness of our suggestions. Two well-known engineering design problems are successfully solved by the proposed methods. © Springer-Verlag Berlin Heidelberg 2013.
Autores principais:	Espírito Santo, I. A. C. P.
Outros Autores:	Costa, Lino; Rocha, Ana Maria A. C.; Azad, Md. Abul Kalam; Fernandes, Edite Manuela da G. P.
Assunto:	Ciências Naturais::Ciências da Computação e da Informação
Ano:	2013
País:	Portugal
Tipo de documento:	capítulo de livro
Tipo de acesso:	acesso aberto
Instituição associada:	Universidade do Minho
Idioma:	inglês
Origem:	RepositóriUM - Universidade do Minho

Descrição
Resumo:	This chapter aims to address the challenging and demanding issue of solving a continuous nonlinear constrained global optimization problem. We propose four stochastic methods that rely on a population of points to diversify the search for a global solution: genetic algorithm, differential evolution, artificial fish swarm algorithm and electromagnetism-like mechanism. The performance of different variants of these algorithms is analyzed using a benchmark set of problems. Three different strategies to handle the equality and inequality constraints of the problem are addressed. An augmented Lagrangian-based technique, the tournament selection based on feasibility and dominance rules, and a strategy based on ranking objective and constraint violation are presented and tested. Numerical experiments are reported showing the effectiveness of our suggestions. Two well-known engineering design problems are successfully solved by the proposed methods. © Springer-Verlag Berlin Heidelberg 2013.