A New Evolutionary Algorithm Based on Decomposition for Multi-Objective Optimization Problems

For multi-objective optimization problems (MOPs), the diversity of obtained solutions is of great importance for multi-objective evolutionary algorithms (MOEAs). To this end, in this paper, a new MOEA based on decomposition of a MOP into a number of constrained single-objective sub-problems (MOEA/D-MCS) is proposed. A MOP is firstly decomposed into a set of simple single-objective optimization sub-problems, and the objective space of the MOP is also decomposed into some number of sub-objective spaces. Then, each sub-objective space is considered as the feasible region of the corresponding single-objective problem. For each sub-problem, solutions of the feasible region are better than solutions of the infeasible region, which can maintain quite well the diversity of obtained solutions. Moreover, MOEA/D-MCS compares with MOEA/D UMOEA/D, GDRE3 d NSGAIII on six multi-objective benchmark functions. Simulation results show that the proposed algorithm is able to obtain a set of solutions with better diversity and convergence.