A MOEA/D with global and local cooperative optimization for complicated bi-objective optimization problems

Multi-objective evolutionary algorithm based on decomposition has been well-recognized in addressing bi-objective optimization problems. However, it is a serious challenge for MOEA/D to optimize real-world complicated bi-objective optimization problems which have irregular optimal objective spaces, such as long-tailed, peaked, and disconnected types. Hence, an improved MOEA/D with global and local cooperative mechanisms is proposed in this paper for complicated bi-objective optimization problems. The overall optimization is carried out by the coordination of global and local search phases. Specifically, a bidirectional global search strategy based on reference points is first used to broaden the search regions. Aiming at the irregularity of the Pareto fronts, an adaptive neighborhood search strategy is proposed according to the individual sparsity, which can further explore near the obtained global optimal solutions. The cooperation of these two strategies is designed to improve the distribution and uniformity of the population on complicated bi-objective optimization problems. The performance of the proposed algorithm is investigated on a representative set of benchmark test functions and a realworld water resource allocation problem. The simulation results show that the proposed algorithm is very promising for complicated bi-objective optimization problems. (c) 2023 Elsevier B.V. All rights reserved.