Machine Learning-Driven Anomaly Detection for Quality Assurance in Recycled Fiber Supply Chains

Authors- Assistant Professor Dr. Pankaj Malik, Muskan Singh, Vasundhara Shukla, Tinkesh Barapatre, Vishesh Mishra

Abstract-In recent years, the demand for recycled fibers in textile manufacturing has surged as industries strive to adopt more sustainable practices. However, maintaining consistent quality in recycled fibers presents significant challenges due to variability in supply chain conditions. Anomalies in transportation, storage, and handling, such as temperature fluctuations and prolonged storage times, can negatively impact fiber quality and, ultimately, the quality of the final textile products. This paper proposes a machine learning-driven approach to anomaly detection across the recycled fiber supply chain, aiming to proactively identify and address quality risks. By leveraging data from environmental sensors, transportation records, and storage logs, various machine learning models—including isolation forests, deep autoencoders, and long short-term memory (LSTM) networks—are developed and evaluated for their effectiveness in detecting supply chain anomalies. Experimental results demonstrate the potential of these models to accurately identify anomalies and provide early warnings, which can inform quality control interventions before production. Case studies highlight specific anomaly scenarios, such as temperature spikes and excessive handling, which were successfully flagged by the models. The study underscores the value of machine learning for real-time quality assurance in recycled fiber supply chains, offering a pathway toward greater consistency and sustainability in textile production. This approach also lays the groundwork for future research integrating Internet of Things (IoT) devices and blockchain for enhanced traceability and accountability in sustainable textile supply chains.

DOI: /10.61463/ijset.vol.12.issue5.298