An Enhanced Direction based Multi-objective Evolutionary Algorithms using Rank Sum

This paper suggests to use rank sum technique to enhance multiobjective evolutionary algorithms (MOEAs), specifically for MOEAs using directional information. We introduce Rank based Converge Direction (RCD) and Rank based Spread Direction (RSD) for DMEA-II (DMEA-II is a popular directional based MOEA), it called DRS. In which, we use rank sum technique in convergence direction and spread direction in place of dominated-non-dominated direction and non-dominated-non-dominated direction. The performance of the proposal is validated and compared with the original algorithm NSGA-II, DMEA-II on well-known benchmark sets. With respect to result, DRS is compared favorably with its origin and competitive with others.