APPLICATION OF GENETIC ALGORITHM TO THE UNRELATED PARALLEL MACHINE SCHEDULING PROBLEM

Abstract

ABSTRACT This research applies the genetic algorithm with elitism strategy to the plastic and chemical industry for the most part and takes Bi-Oxide Polypropylene (BOPP) FILM process as an example. Scheduling of BOPP FILM belongs to unrelated parallel machine scheduling problem. It considers not only the sequencing but also dispatching the orders to machines. The past literatures about unrelated parallel machine concentrate on the optimal solutions with respect to single objective. But, reviewing the real world case, single-objective scheduling problems have not characterized the critical production environment well. Instead of the single-objective scheduling problems is the multi-objective ones. The objectives are usually competitive owing to the limited resource. Therefore how to optimize among the competitive objectives is the key issue in this research. The objective is the minimization of the sum of the waste material cost and makespan in this research. In this research, we propose the dispatching rule “Earliest Finished First Dispatched (EFFD)” to reduce the makespan. The optimal levels of parameters in the genetic algorithm are settled through Taguchi method. We evaluate the efficiency of our model via Lingo 6.0 with respective to the small size problem. As the number of orders increases, Lingo 6.0 cannot timely provide the optimal solutions. Finally, the performance of genetic algorithm with elitism strategy is compared with the traditional dispatching rules, EDD, SPT, LPT, and Tabu search. All the comparisons result in genetic algorithm with elitism is superior to the other approaches and the percent improvement is around 8. Consequently the empirical results are deserved to be mentioned in the real world case.