Modified artificial fish swarm algorithm to solve unrelated parallel machine scheduling problem under fuzzy environment

Unrelated parallel machine scheduling problem (UPMSP) in a fuzzy environment is an active research area due to the fuzzy nature of most real-world problems. UPMSP is an NP-hard problem; thus, finding optimal solutions is challenging, particularly when multiple objectives need to be considered. Hence...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:AIMS mathematics Ročník 9; číslo 12; s. 35326 - 35354
Hlavní autoři: Ibadi, Azhar Mahdi, Rahman, Rosshairy Abd
Médium: Journal Article
Jazyk:angličtina
Vydáno: AIMS Press 01.01.2024
Témata:
artificial fish swarm algorithm
fuzzy sets
modified algorithm
multi-objective
unrelated parallel machine scheduling problem
ISSN:2473-6988, 2473-6988
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Unrelated parallel machine scheduling problem (UPMSP) in a fuzzy environment is an active research area due to the fuzzy nature of most real-world problems. UPMSP is an NP-hard problem; thus, finding optimal solutions is challenging, particularly when multiple objectives need to be considered. Hence, a metaheuristic algorithm based on a modified artificial fish swarm algorithm (AFSA) is presented in this study to minimize the multi-objective makespan and total tardiness. Three modifications were made to the proposed algorithm. First, aspiration behavior was added to AFSA behaviors to increase effectiveness. Second, improved parameters such as step and visual were used to balance global search capability and convergence rate. Finally, a transformation method was injected to make the algorithm suitable for discrete optimization problems such as UPMSP. The proposed algorithm was compared with AFSA and five modified versions of AFSA to verify and measure the algorithm's effectiveness by conducting three different sizes of problems. Afterward, the Wilcoxon signed-rank test was used to statistically evaluate the algorithm's performance. The results indicate that the proposed algorithm significantly outperformed the other algorithms, especially for medium and large-sized problems.
ISSN:2473-6988
2473-6988
DOI:10.3934/math.20241679