Research and Optimization of a Real-Time Quality Monitoring System for Smart Production Lines Based on IoT Sensors

Abstract

The rise of Industry 4.0 has accelerated the adoption of IoT-enabled sensing for real-time quality assurance in smart manufacturing. However, most existing systems depend on cloud-centric analytics or supervised learning models that require extensive labeled defect data, leading to latency, poor adaptability, and limited applicability in dynamic production environments. To address this gap, this study proposes an IoT sensor-driven quality monitoring framework based on multi-modal signal acquisition, edge computing, and adaptive thresholding informed by short-term process variability. The system was deployed and evaluated on two industrial production lines, Bosch automotive components and CATL lithium-ion module assembly, using longitudinal tracking of defect rate, first-pass yield, and overall equipment effectiveness. Results indicate a 26-31% reduction in defects, a 26.9% increase in first-pass yield, and a 7% improvement in OEE, alongside a 47.8% decrease in false alarms compared with static control methods. These findings demonstrate that real-time adaptive monitoring can enhance quality performance without dependency on large labeled datasets. The study provides a replicable implementation methodology and insights into sensor contribution, offering practical guidance for scalable deployment and future advancements in intelligent quality control.