Early prediction of fabric quality using machine learning to reduce rework in manufacturing processes

• Published in: An international journal of optimization and control Vol. 14; no. 4; pp. 308 - 321
• Main Authors: Aydın, Sema, Altun, Koray
• Format: Journal Article
• Language: English
• Published: Balikesir University 09.10.2024
• Subjects: Artificial neural networks; Fabric quality; machine learning; Rework reduction
• ISSN: 2146-0957; 2146-5703
• DOI: 10.11121/ijocta.1462

The increasing competition and rapid technological advancements in today's business world have raised customer expectations. People now expect quick delivery, low prices, and high-quality products. As a result, companies must adapt to this competitive environment to survive. Rework, which is a significant cost in production, increases expenses, reduces production efficiency, and can lead to customer attrition. Research shows various efforts across different sectors to reduce rework, although there is still a gap in the textile sector's fabric dyeing units. Common problems in these units include non-retentive colors, customer dissatisfaction with shades, and repeated dyeing due to environmental factors or dye vat issues. This study uses logistic regression and artificial neural networks models from machine learning to predict which fabrics will need rework, using data from a textile company in Bursa. The analysis indicates that artificial neural networks models perform better.