PENGEMBANGAN ALGORITMA SIMULATED ANNEALING UNTUK LINTASAN PERAKITAN DENGAN MEMPERTIMBANGKAN JENIS TOOLS DAN WAKTU SETUP PADA KOLABORASI MANUSIA-ROBOT

• Main Author: Rachmana Rahman, Ardhiefa
• Format: Theses
• Language: Indonesia
• Subjects: Teknik (Rekayasa, enjinering dan kegiatan berkaitan)
• Online Access: https://digilib.itb.ac.id/gdl/view/81194
• Institution: Institut Teknologi Bandung

In the Assembly Line Balancing Problem (ALBP) on the assembly line, cycle time and the lack of operator choice on the assembly line are a concern. The smaller the cycle time, the faster the product delivery process to customers. Currently, a mathematical model has been developed related to minimizing cycle time by utilizing Human-Robot Collaboration along with consideration of installing types of tools on the cobot end-effector. The weakness of mathematical or analytical models is that their computing time is very long. This long computing time will obstruct the assembly line reconfiguration process. To solve this computing time problem, in this research a Simulated Annealing (SA) algorithm was developed. The SA algorithm being developed consists of two major algorithms, namely: the construction algorithm to produce an initial solution, and the SA improvement algorithm which consists of the outer loop algorithm which is the general procedure for the SA algorithm and the inner loop algorithm as a generator of new solutions. The inner loop procedure consists of three procedures, namely task transfer, task exchange and resource exchange. The other contribution that developed there is also the addition of the beta (?) parameter as a parameter to limit the solution space in the form of upper bound cycle time. This research uses a design of experiment (DOE) in the form of a full factorial design to determine the Simulated Annealing parameters that significantly influence cycle time. Apart from that, the ? parameter is also adjusted so that the best combination of parameters will be obtained which will be used in the SA algorithm test. From the experimental design that has been carried out, there are two SA parameters that significantly influence cycle time. These two parameters are the number of temperatur drops (M) and the number of iterations at each temperature (N). The parameter setting ? will be optimal when ? = 0.7 for small data and ? = 0.9 for large data. Based on testing the SA algorithm that has been developed, the results obtained are: an improvement in the new solution of 11.38% from the initial solution, the difference the cycle time gap between the analytical method and the SA algorithm is only 1.67%, as well as a very significant reduction in computing time between the analytical method and the SA algorithm, namely 88.78%. Looking at the percentage of results from the SA algorithm, it is evident that the SA algorithm developed produces solution quality that is close to optimal and computing time speed is better than the analytical method.