A new niching method for the direction-based multi-objective evolutionary algorithm

The direction of improvement has been discussed and used to guide MOEAs during the search process towards the area of Pareto optimal set. One of typical examples using direction of improvement is the Direction based Multi-objective Evolutionary Algorithm (DMEA). For DMEA, its authors introduced a novel algorithm incorporating the concept of direction of improvement. Our preliminary analysis showed that the performance of DMEA is also dependent on the way niching is implemented. In this paper, we propose a new niching approach for DMEA. The main idea of proposed approach is to define a new concept of ray-based density within the framework of DMEA and then use it as niching information. With this method, we hope to give more control on the balance between exploration and exploitation. To validate the performance of the new improved version of DMEA, we carried out a case study on several test problems and comparison with some other MOEAs, it obtained quite good results on primary performance metrics, namely the generation distance, inverse generation distance and hypervolume.