Authors:
Alexey Vakhnin
and
Evgenii Sopov
Affiliation:
Department of System Analysis and Operations Research, Reshetnev Siberian State University of Science and Technology, Krasnoyarsk and Russia
Keyword(s):
Large-scale Global Optimization, Variable Grouping Method, Cooperative Coevolution, Evolutionary Algorithms.
Related
Ontology
Subjects/Areas/Topics:
Evolutionary Computation and Control
;
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Optimization Algorithms
Abstract:
Large-scale global optimization (LSGO) is known as one of the most challenging problem for evolutionary algorithms (EA). In this study, we have proposed a novel method of grouping variables for the cooperative coevolution (CC) framework (random adaptive grouping (RAG))). We have implemented the proposed approach in a new evolutionary algorithm (DECC-RAG), which uses the Self-adaptive Differential Evolution (DE) with Neighborhood Search (SaNSDE) as the core search technique. The RAG method is based on the following idea: after some predefined number of fitness evaluations in cooperative coevolution, a half of subcomponents with the worst fitness values randomly mixes indices of variables, and the corresponding evolutionary algorithms reset adaptation of parameters. We have evaluated the performance of the DECC-RAG algorithm with the large-scale global optimization (LSGO) benchmark problems proposed within the IEEE CEC 2010. The results of numerical experiments are presented and discus
sed. The results have shown that the proposed algorithm outperforms some popular LSGO approaches.
(More)