Constraint-activated differential evolution for constrained min-max optimization problems

Theory and methodology

Shu-Mei Guo, Chin Chang Yang, Hsin Yu Chang, Jason Sheng-Hon Tsai

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

A constraint-activated differential evolution is proposed to solve constrained min-max optimization problems in this paper. To provide theoretical understanding for these problems, their global optima are specified in the proposed definitions. Based on the definition, we propose theorems to prove that a min-max algorithm can be used to solve a max-min problem without any algorithmic changes. Based on the theorems, we propose a constraint-activated differential evolution to solve constrained min-max problems. The proposed method consists of three components, propagation, constraint activation, and inner level evolution. The propagation provides exploitation power of evolution. The constraint activation directly finds a solution which can best activate constraints. The inner level evolution provides continuous evolutionary behavior to prevent convergence premature. The simulation results show that the proposed method attains 100% success rates for all of the numerical benchmarks with an exploitative mutation strategy.

Original languageEnglish
Pages (from-to)1626-1636
Number of pages11
JournalExpert Systems With Applications
Volume42
Issue number3
DOIs
Publication statusPublished - 2015 Feb 15

Fingerprint

Chemical activation

All Science Journal Classification (ASJC) codes

  • Engineering(all)
  • Computer Science Applications
  • Artificial Intelligence

Cite this

@article{15d8471d61c24ab8ac7198b21ac13663,
title = "Constraint-activated differential evolution for constrained min-max optimization problems: Theory and methodology",
abstract = "A constraint-activated differential evolution is proposed to solve constrained min-max optimization problems in this paper. To provide theoretical understanding for these problems, their global optima are specified in the proposed definitions. Based on the definition, we propose theorems to prove that a min-max algorithm can be used to solve a max-min problem without any algorithmic changes. Based on the theorems, we propose a constraint-activated differential evolution to solve constrained min-max problems. The proposed method consists of three components, propagation, constraint activation, and inner level evolution. The propagation provides exploitation power of evolution. The constraint activation directly finds a solution which can best activate constraints. The inner level evolution provides continuous evolutionary behavior to prevent convergence premature. The simulation results show that the proposed method attains 100{\%} success rates for all of the numerical benchmarks with an exploitative mutation strategy.",
author = "Shu-Mei Guo and Yang, {Chin Chang} and Chang, {Hsin Yu} and Tsai, {Jason Sheng-Hon}",
year = "2015",
month = "2",
day = "15",
doi = "10.1016/j.eswa.2014.09.051",
language = "English",
volume = "42",
pages = "1626--1636",
journal = "Expert Systems with Applications",
issn = "0957-4174",
publisher = "Elsevier Limited",
number = "3",

}

Constraint-activated differential evolution for constrained min-max optimization problems : Theory and methodology. / Guo, Shu-Mei; Yang, Chin Chang; Chang, Hsin Yu; Tsai, Jason Sheng-Hon.

In: Expert Systems With Applications, Vol. 42, No. 3, 15.02.2015, p. 1626-1636.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Constraint-activated differential evolution for constrained min-max optimization problems

T2 - Theory and methodology

AU - Guo, Shu-Mei

AU - Yang, Chin Chang

AU - Chang, Hsin Yu

AU - Tsai, Jason Sheng-Hon

PY - 2015/2/15

Y1 - 2015/2/15

N2 - A constraint-activated differential evolution is proposed to solve constrained min-max optimization problems in this paper. To provide theoretical understanding for these problems, their global optima are specified in the proposed definitions. Based on the definition, we propose theorems to prove that a min-max algorithm can be used to solve a max-min problem without any algorithmic changes. Based on the theorems, we propose a constraint-activated differential evolution to solve constrained min-max problems. The proposed method consists of three components, propagation, constraint activation, and inner level evolution. The propagation provides exploitation power of evolution. The constraint activation directly finds a solution which can best activate constraints. The inner level evolution provides continuous evolutionary behavior to prevent convergence premature. The simulation results show that the proposed method attains 100% success rates for all of the numerical benchmarks with an exploitative mutation strategy.

AB - A constraint-activated differential evolution is proposed to solve constrained min-max optimization problems in this paper. To provide theoretical understanding for these problems, their global optima are specified in the proposed definitions. Based on the definition, we propose theorems to prove that a min-max algorithm can be used to solve a max-min problem without any algorithmic changes. Based on the theorems, we propose a constraint-activated differential evolution to solve constrained min-max problems. The proposed method consists of three components, propagation, constraint activation, and inner level evolution. The propagation provides exploitation power of evolution. The constraint activation directly finds a solution which can best activate constraints. The inner level evolution provides continuous evolutionary behavior to prevent convergence premature. The simulation results show that the proposed method attains 100% success rates for all of the numerical benchmarks with an exploitative mutation strategy.

UR - http://www.scopus.com/inward/record.url?scp=84908519368&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84908519368&partnerID=8YFLogxK

U2 - 10.1016/j.eswa.2014.09.051

DO - 10.1016/j.eswa.2014.09.051

M3 - Article

VL - 42

SP - 1626

EP - 1636

JO - Expert Systems with Applications

JF - Expert Systems with Applications

SN - 0957-4174

IS - 3

ER -