A Machine Learning Framework for Root Cause Analysis in ISO-Compliant Manufacturing Using RFID Data

This study proposes a scalable machine learning (ML) framework for automated root cause analysis (RCA) in ISO 9001-compliant manufacturing environments using real-time Radio Frequency Identification (RFID) data. The framework integrates heterogeneous data streams from RFID-enabled production systems with supervised and unsupervised learning models to detect anomalies and localize their underlying causes. A structured pipeline encompassing data acquisition, preprocessing, feature engineering, anomaly detection, and explainable RCA are developed to address challenges associated with noisy, incomplete, and high-velocity industrial data. To ensure alignment with ISO 9001 quality management principles, the framework incorporates traceability, documented decision-making, and continuous improvement mechanisms within a Plan–Do–Check–Act (PDCA) cycle. Experimental validation using a synthetic industrial dataset demonstrates that the proposed approach reduces mean time to root cause identification by over 35% and improves defect detection accuracy (F1-score > 0.90) compared to conventional rule-based methods. The study further introduces explainable AI (XAI) techniques to enhance interpretability and auditability of ML outputs, thereby supporting compliance requirements. The results highlight the practical viability of integrating RFID-driven analytics with ML to enable predictive quality control, reduce operational inefficiencies, and advance Industry 4.0 adoption. This research contributes a structured, ISO-aligned framework that bridges the gap between advanced analytics and real-world manufacturing quality systems.