Multi-objective genetic algorithm based on learning effect for solving hybrid flow shop scheduling problem

Scheduling is one of the most important issues that enable companies to survive in production environments where competition is very intense with the development of technology. In this study, the production system of a company producing emergency lighting products was discussed and it was determined that the scheduling type belonged to the Hybrid Flow Shop Scheduling (HFSS) problem class. In addition, the speed and quality of the work performed in this manufacturing environment are largely based on the manual skills of the operators. In this study, a mixed integer linear mathematical model for solving the multi-objective HFSS problem, which considers the setup times between jobs, removal times, and learning effect, is proposed to minimize the maximum completion time and total tardiness. Due to the complex structure of the problem and its multi-objective nature, a Multi-Objective Genetic Algorithm (MOGA) solution approach based on three different scalarization methods, Weighted-Sum (WSM), Conic (CSM) and Tchebycheff (TSM), has been proposed. The performance of the scalarization variant GA was tested using two different performance metrics: weighted distance to the ideal point and weighted distance to the reference point, with 21 benchmark instances taken from the literature, small, medium, and large. The parameter settings of the proposed MOGA were determined using irace. The comparison results show that GA with WSM performs better than other algorithms in terms of the weighted distance to the ideal point, while GA with TSM performs better than GA with CSM in accordance with the weighted distance to the reference point. The proposed algorithm was applied to the company. According to the experimental results, a Decision Support System with a user-friendly interface has been designed, through which the suggested algorithm can work effectively, and the company can use it.