A Novel Hybrid Multi-objective Optimization Framework: Rotating the Objective Space


Multi-objective Evolutionary Algorithms (MOEAs) are popular ap- proaches for solving multi-objective pr oblems (MOPs). One representative me- thod is Non-dominated Sorting Genetic Algorithm II (NSGA-II), which has achieved great success in the field by introducing non-dominated sorting into survival selection. However, as a common issue for dominance-based algo- rithms, the performance of NSGA-II will decline in solving problems with 3 or more objectives. This paper aims to circumvent this issue by incorporating the concept of decomposition into NSGA-II. A grouping-based hybrid multi- objective optimization framework is proposed for tackling 3-objective prob- lems. Original MOP is decomposed into several scalar subproblems, and each group of population is assigned with two scalar subproblems as new objectives. In order to better cover the whole objective space, new objective spaces are formulated via rotating the original objective space. Simulation results show that the performance of the proposed algorithm is competitive when dealing with 3-objective problems.