GD-MOEA: A New Multi-Objective Evolutionary Algorithm Based on the Generational Distance Indicator

Abstract

In this paper, we propose a new selection mechanism for Multi-Objective Evolutionary Algorithms (MOEAs), which is based on the generational distance indicator and uses a technique that relies on Euclidean distances to maintain diversity in the population (in objective function space). Our proposed selecion mechanism is incorporated into a MOEA which adopts the operators of NSGA-II (crossover and muta- tion) to generate new individuals. The new MOEA is called Generational Distance - Multi-Objective Evolutionary Algorithm (GD-MOEA). Our GD-MOEA is validated using standard test problems taken from the spe- cialized literature, having three to six objective functions. GD-MOEA is compared with respect to MOEA/D using Penalty Boundary Intersection (PBI), which is based on decomposition, and to SMS-EMOA-HYPE (a version of SMS-EMOA that uses a fitness assignment scheme based on the use of an approximation of the hypervolume indicator). Our prelimi- nary results indicate that if we consider both quality in the solutions and the running time required to generate them, our GD-MOEA is a good alternative to solve multi-objective optimization problems having both low dimensionality and high dimensionality in objective function space.