Title: Non-dominated sorting genetic algorithm with decomposition to solve constrained optimisation problems
Authors: Sanyou Zeng; Dong Zhou; Hui Li
Addresses: School of Computer Science, China University of GeoSciences, 430074 Wuhan, Hubei, China ' School of Computer Science, China University of GeoSciences, 430074 Wuhan, Hubei, China ' School of Computer Science, China University of GeoSciences, 430074 Wuhan, Hubei, China
Abstract: Pareto-domination was adopted to handle not only trade-off between objective and constraints but also trade-off between convergence and diversity on solving a constrained optimisation problem (COP) in this paper like many other researchers. But there are some differences. This paper converts a COP into an equivalent dynamic constrained multi-objective optimisation problem (DCMOP) first, then dynamic version of non-dominated sorting genetic algorithm with decomposition (NSGA/D) is designed to solve the equivalent DCMOP, consequently solve the COP. A key issue for the NSGA/D working effectively is that the environmental change should not destroy the feasibility of the population. With a feasible population, the NSGA/D could solve well the DCMOP just as a MOEA usually can solve well an unconstrained MOP. Experimental results show that the NSGA/D outperforms or performs similarly to other state-of-the-art algorithms referred to in this paper, especially in global search.
Keywords: evolutionary algorithms; constrained optimisation; multi-objective optimisation; dynamic optimisation; non-dominated sorting genetic algorithms; NSGA; decomposition.
DOI: 10.1504/IJBIC.2013.055080
International Journal of Bio-Inspired Computation, 2013 Vol.5 No.3, pp.150 - 163
Received: 30 Aug 2012
Accepted: 03 Sep 2012
Published online: 31 Mar 2014 *