Opposition -based genetic algorithm approach to flexible job-shop scheduling problem

Abstract

Production scheduling, or flexible job-shop scheduling problem in specific, has widely been a regular practice for most manufacturing firms, it is necessary for businesses to come up with the best schedules in the most time-efficient manner. An effective schedule will minimize the amount of unnecessary time for production or operation, increase overall efficiency, and consequently reduce the costs incurred. The proposed algorithms in this study integrated the workflow and concepts from conventional genetic algorithm (GA), and opposition-based learning (OBL) to improve the quality of chromosome population, eventually enhance the overall converging performance. To further tackle the issue of premature convergence, a multiple-restart (MR) strategy, as well as a diversity verification scheme using Shannon’s diversity index were utilized to provide additional support for algorithms to escape from the local traps. The proposed algorithms were tested on several infamous benchmark instances to record their performance and solution deviation. And finally, a sensitivity analysis regarding a few parameters were also performed in order to validate whether there is any correlation between these parameters and the final accuracy.