Hibridização de algoritmos genéticos com sistemas imunológicos artificiais para problemas de otimização com restrições em engenharia

Abstract

In this work some possibilities of hibridization of genetic algorithms and artificial immune systems are proposed, computationally implemented, and tested in order to solve constrained optimization problems. Such problems arise frequently in the computational modeling of relevant scientific and technological problems. After a brief formulation and solution characterization for a constrained optimization problem in Ɽn, an introduction is made to structural modeling and the optimization problems in this area, showing their relevance and the dificulties arising in the development of approximate solution methods. Several bio-inspired meta-heuristics for solving such optimization problems are then presented with emphasis on genetic algorithms (GAs) and artiticial immune systems (AISs). Some possible GA-SIA hybrids are then proposed. In order to assess the performance of the proposed procedures, computational experiments are conducted considering several test-problems from the literature and, finally, conclusions as well as future work proposals are presented.