A modified MOEA/D by introducing an invasive weed optimization operator

Abstract

Multi-objective Evolutionary Algorithm Based on Decomposition (MOEA/D) is a representative multi-objective optimization (MOO) algorithm. Recently, some researchers proposed several modification versions of MOEA/D to improve the performance on the benchmark test problems suggested in CEC'09. This paper proposes a modified invasive weed optimization (IWO) operator and introduced it into two modification versions of MOEA/D. When the generation is divisible by certain number, the modified IWO operator rather than Differential Evolutionary (DE) operator is implemented to generate new individuals. And updating the individuals by using the modified IWO operator has a little different from the original IWO operator, the new individuals will be compared with their parent individuals firstly, then perform the original update method. Experimental results of 10 unconstrained problems proposed in CEC'09 show that the performance of the proposed algorithm has an improvement for some of the test problems.