원문정보
초록
영어
DE is a topic of current interest in the optimization field. It is the most capable evolutionary algorithm based on biological theory of evolution because of its ease and competence in solving variety of problems, like multi-objective, multi-modal, dynamic optimization problems. But premature convergence or stagnation is a main problem with it. So In order to improve the performance of DE, significant number of DE variants has been proposed by many researchers over the last few decades. Mutation is one of the key tasks of DE. It appreciably influences the performance of DE. In this paper, DE variants with four different mutation techniques- DE/rand/1, DE/local-to-best, DE/either-or and MODE are studied and implemented. Comparison of DE having these mutation strategies is made for variety of dimension and population size and results shows that DE/local-to-best performs best on all the benchmark functions where as MODE also show significant performance.
목차
1. Introduction
2. Differential Evolution
2.1 Outline of Differential Evolution
2.2 Mutation Strategies
3. Implementation Model
4. Experimental Analysis
4.1 Evaluation on the Basis of Minimum Cost
4.2 Evaluation on the Basis of Convergence Time
4.3 Evaluation on the Basis of Number of Functions Evaluated
5. Conclusion and Future Scope
References