Integrating preferences within multiobjective flexible job shop scheduling

Abstract

When faced with a multiobjective optimization problem, it is necessary to consider the decision-maker preferences in order to propose the best compromise solution. We consider the multiobjective flexible job shop scheduling problem and a decision-maker that is best represented using a non-compensatory reference level-based preference model. We show how integrating this model into a multiobjective genetic algorithm allows to obtain solutions that surpass more aspiration levels when compared to classical multiobjective optimization approaches. Furthermore, these solutions are found faster and in greater numbers which facilitates their integration within the workshop.