Constrained multi-objective optimization evolutionary algorithm

Abstract

Genetic algorithms for constrained multi-objective optimization problems mainly focus on optimizing the conflicting multiple objectives without considering the constraint conditions. This paper describes a genetic algorithm which uses neighborhood comparisons and archiving in the genetic algorithm to smooth the conflicting objectives. Infeasibility degree selection is used to handle the constraints with the constraint domain principle applied to guide the evolutionary process. Two classic difficult problems constrained multi-objective optimization were analyzed by the algorithm to show that the method can find feasible Pareto solutions with a large probability.