A Multi-objective Evolutionary Algorithm Based on Cone Heart Point Adjustment Vectors to Solve Concave-convex Problems

Decomposition-based multi-objective optimization algorithms have been shown to be successful in solving multi-objective optimization problems (MOPs). Since the shape is unknown before the search, the classical decomposition method relies on a set of uniformly distributed reference vectors to divide the objective space into multiple subregions. However, the the distribution of reference vectors produced by the Das and Dennis systematic approach is difficult to maintain the diversity of (extremely) concave and (extremely) convex problems. To remedy this issue, a multi-objective evolutionary algorithm based on cone heart point adjustment vectors is proposed to solve concave and convex problems (CHP-MOEA). In this paper, the concavity or convexity is first according to the intermediate objective vector. Then, the distribution of reference vectors are adjusted according to the cone heart point on the basis of the traditional vectors. Finally, experimental results on 31 benchmark problems show that the proposed algorithm exhibits better performance than other state-of-the-art algorithms.