A constraint programming-based genetic algorithm for capacity output optimization
Purpose: The manuscript presents an investigation into a constraint programming-based genetic algorithm for capacity output optimization in a back-end semiconductor manufacturing company. Design/methodology/approach: In the first stage, constraint programming defining the relationships between varia...
Gespeichert in:
Veröffentlicht in: | Journal of industrial engineering and management 2014-01, Vol.7 (5), p.1222-1249 |
---|---|
Hauptverfasser: | , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | Purpose: The manuscript presents an investigation into a constraint programming-based genetic algorithm for capacity output optimization in a back-end semiconductor manufacturing company. Design/methodology/approach: In the first stage, constraint programming defining the relationships between variables was formulated into the objective function. A genetic algorithm model was created in the second stage to optimize capacity output. Three demand scenarios were applied to test the robustness of the proposed algorithm. Findings: CPGA improved both the machine utilization and capacity output once the minimum requirements of a demand scenario were fulfilled. Capacity outputs of the three scenarios were improved by 157%, 7%, and 69%, respectively. Research limitations/implications: The work relates to aggregate planning of machine capacity in a single case study. The constraints and constructed scenarios were therefore industry-specific. Practical implications: Capacity planning in a semiconductor manufacturing facility need to consider multiple mutually influenced constraints in resource availability, process flow and product demand. The findings prove that CPGA is a practical and an efficient alternative to optimize the capacity output and to allow the company to review its capacity with quick feedback. Originality/value: The work integrates two contemporary computational methods for a real industry application conventionally reliant on human judgement. |
---|---|
ISSN: | 2013-0953 2013-8423 2013-0953 |
DOI: | 10.3926/jiem.1070 |