ADAPTASI ALGORITMA GENETIKA UNTUK MASALAH PENJADWALAN ULANG DENGAN TEKNIK SISIPAN PADA SISTEM MANUFAKTUR JOB SHOP DINAMIS NON DETERMINISTIK

Abstract

The scheduling of a non-deterministic dynamic job shop was a matter of orga-nizing an amount of operations for certain machines with different sequence for different job, whereas the output of the job varies and was previously could not be determined. The purpose of the research was to minimize the mean tardiness of lateness of a job whether it was either an active job or non-delay, ahead of schedule or the rescheduled, and a prioritized job or the non priority. The research used genetic algorithm approach to determine the problems occurred in the scheduling or the rescheduling of a non-deter-ministic dynamic job shop. The function of fitness of the research functioned as the goal, whereas the best fitness output was the fitness with the biggest minimum mean tardiness. The best fitness for an ahead of schedule, either active or non-delay, was the ones with the exact fitness value of 0,0256410The same goes for early scheduling, either with or without priority, which had the exact output of 0,0256410. For a new job, either active or non-delay, had the exact fitness output of 0, 0148148. A different resulted for the resche-duling of a prioritized new job that had the best fitness output of 0, 0077295 for active job; and 0,005868 for a non-delay job. Rescheduling could be done by doing a job insertion in the beginning of the schedule. It was seen that in the beginning of the schedule, whether active or non-delay, the fitness was 0,0245610 and afterwards by adding job 6, job 7 and job 8 the output of the fitness differed accordingly with the case of the job.