A guided artificial bee colony (GABC) heuristic for permutation flowshop scheduling problem (PFSP)

• Main Author: Sidek, Noor Azizah
• Format: Thesis
• Language: English; English; English
• Published: 2021
• Subjects: TS155-194 Production management. Operations management
• Online Access: http://eprints.uthm.edu.my/1832/2/NOOR%20AZIZAH%20BINTI%20SIDEK%20-%20declaration.pdf; http://eprints.uthm.edu.my/1832/1/NOOR%20AZIZAH%20BINTI%20SIDEK%20-%2024p.pdf; http://eprints.uthm.edu.my/1832/3/NOOR%20AZIZAH%20BINTI%20SIDEK%20-%20full%20text.pdf; http://eprints.uthm.edu.my/1832/

Flowshop is the most common production system in the industry, and there are many documented efforts to improve the performance of the flowshop. The range spreads from the usage of heuristics to metaheuristics, and one of the promising methods is NEH (Nawaz, Enscore & Ham) heuristics. This study aims to improve NEH, using an enhanced version of Artificial Bee Colony (ABC) algorithm because the original one has the problem of slow converge speed. As a result, this study will propose a mechanism to improve the convergence speed of ABC because faster convergence speed is the ability to find high-quality results in lesser iterations compared to others. The study clusters the Employed Bees (EB) and Onlooker Bees (OB) into several groups: Total Greedy, Semi Greedy and Non-Greedy. Upon completion, the study selected the Total Greedy (3+0+0) because of the leading performance in makespan value (performance indicator), and the author used it for the rest of this study. This study proposed two variants of the guided initial ABC or Guided Artificial Bee Colony (GABC) with one variant (NEH-based ABC), employing the concept of NEH and the second variant (GABC), employing the concept of NEH and First Job Sequence Arrangement Method. The study experimented according to ten datasets of Taillard benchmark and divided the experiments into several categories and the experiments run every data for several iterations, and for each dataset, there are 20 replications. This study compared the performance of NEH, ABC, NEH-based ABC and GABC, which also act as the validation process. Based on the results, ABC produced inconsistent results for a significant amount of times and interestingly, GABC, NEHbased ABC and ABC produced 68.75%, 63.33% and 0.01% results that are better than NEH, respectively. The data also shows that GABC is 37.9% better than its variant. Finally, the author can conclude that this study demonstrated the slow convergence issue of ABC.