Difficulty Controllable and Scalable Constrained Multi-objective Test Problems

In this paper, we propose a general toolkit to construct constrained multi-objective optimisation problems (CMOPs) with three different kinds of constraint functions. Based on this toolkit, we suggested eight constrained multi-objective optimisation problems named CMOP1-CMOP8. As the ratio of feasib...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2015 International Conference on Industrial Informatics - Computing Technology, Intelligent Technology, Industrial Information Integration S. 76 - 83
Hauptverfasser: Zhun Fan, Wenji Li, Xinye Cai, Hui Li, Kaiwen Hu, Haibin Yin
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.12.2015
Schlagworte:
Aerospace electronics
Computer science
Constrained Multi-objective Evolutionary Algorithm
Constrained Multi-objective Optimisation problem
Evolutionary computation
Linear programming
Pareto optimization
Shape
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In this paper, we propose a general toolkit to construct constrained multi-objective optimisation problems (CMOPs) with three different kinds of constraint functions. Based on this toolkit, we suggested eight constrained multi-objective optimisation problems named CMOP1-CMOP8. As the ratio of feasible regions in the whole search space determines the difficulty of a constrained multi-objective optimisation problem, we propose four test instances CMOP3-6, which have very low ratio of feasible regions. To study the difficulties of proposed test instances, we make some experiments with two popular CMOEAs - MOEA/D-CDP and NSGA-II-CDP, and analysed their performances.
DOI:10.1109/ICIICII.2015.105