A constrained multi-objective evolutionary algorithm assisted by an additional objective function

In constrained multi-objective optimization, the degree of constraint violation as an additional objective function has been optimized together with the original M objective functions for better diversity. However, it still faces the challenge of deeply exploring feasible regions while maintaining t...

Full description

Saved in:
Bibliographic Details
Published in:Applied soft computing Vol. 132; p. 109904
Main Authors: Yang, Yongkuan, Huang, Pei-Qiu, Kong, Xiangsong, Zhao, Jing
Format: Journal Article
Language:English
Published: Elsevier B.V 01.01.2023
Subjects:
Additional objective function
Constrained optimization
Evolutionary algorithms
Feasible regions
Multi-objective optimization
Feasible regions
Multi-objective optimization
Evolutionary algorithms
Additional objective function
Constrained optimization
ISSN:1568-4946
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In constrained multi-objective optimization, the degree of constraint violation as an additional objective function has been optimized together with the original M objective functions for better diversity. However, it still faces the challenge of deeply exploring feasible regions while maintaining the diversity of the population. To this end, this paper proposes a novel constrained multi-objective evolutionary algorithm assisted by an additional objective function, called CMAOO. First, the main population is constructed to optimize an (M+1)-objective optimization problem consisting of the original M objective functions and the degree of constraint violation. Additionally, all the feasible solutions are saved in an external archive. Then, the main population and the external archive are evolved to search the whole space and the feasible regions, respectively. After that, their offspring are combined to separately update the external archive and the main population. Experimental studies are conducted to test the performance of CMAOO with four state-of-the-art algorithms on 34 test problems and a real-world problem. The results demonstrate that CMAOO is competitive to solve constrained multi-objective optimization problems. •The main population solves an (M + 1)-objective optimization problem.•An external archive maintains the obtained feasible solutions..•The offspring updates the external archive and the main population.
ISSN:1568-4946
DOI:10.1016/j.asoc.2022.109904