An improved MOEA/D for multi-objective job shop scheduling problem

Abstract

In this article, an improved multi-objective evolutionary algorithm, which is based on decomposition (IMOEA/D) for multi-objective job shop scheduling problem, is proposed to solve multiple objectives job shop scheduling problems. Three minimisation objectives – the maximum completion time (makespan), the total flow time and the tardiness time are considered simultaneously. In the proposed algorithm, several prior rules are presented to construct the initial population with a high level of quality. Meanwhile, according to the contribution of each operator to the external archive, an adaptive mechanism is adopted to select corresponding operators to generate new solutions, which can accelerate convergence speed. Simulation results on the standard test instances show that IMOEA/D has a better convergence performance compared with multi-objective evolutionary algorithms based on Pareto dominance.