In Situ Surface Defect Detection in Polymer Tube Extrusion: AI-Based Real-Time Monitoring Approach

Abstract

While striving to optimize overall efficiency, smart manufacturing systems face various problems presented by the aging workforce in modern society. The proportion of aging workers is rapidly increasing worldwide, and visual perception, which plays a key role in quality control, is significantly susceptible to the impact of aging. Thus it is necessary to understand these changes and implement state-of-the-art technologies as solutions. In this study, we conduct research to mitigate the negative effects of aging on visual recognition through the synergistic effects of real-time monitoring technology combining cameras and AI in polymer tube production. Cameras positioned strategically and with sophisticated AI within the manufacturing environment promote real-time defect detection and identification, enabling an immediate response. An immediate response to defects minimizes facility downtime and enhances the productivity of manufacturing industries. With excellent detection performance (approximately 99.24%) and speed (approximately 20 ms), simultaneous defects in a tube can be accurately detected in real time. Finally, real-time monitoring technology with adaptive features and superior performance can mitigate the negative impact of decreased visual perception in aging workers and is expected to improve quality consistency and quality management efficiency.