Revolutionizing the Application of Automatic Inspection System for Industrial Parts Using AI Machine Vision Technology
Baojun Wang,Yulan Lv
DOI: https://doi.org/10.12694/scpe.v25i4.2083
2024-06-16
Abstract:This paper addresses the challenges of large data management, prolonged operation times, and low detectionefficiency encountered in automatic detection systems. To overcome these issues, we propose a novel research and application method utilizing AI machine vision technology. The methodology employs the Pulse-Coupled Neural Network (PCNN) algorithm for analyzing machine control points, enhancing system reliability and detection efficiency. Furthermore, a three-stage sliding table mechanism is implemented to facilitate seamless operation restarts. Notably, our approach significantly reduces the time required for key operations, such as feeding, imaging, decision-making, on-site inspection, and re-inspection, all within 5 seconds and 1 meter distance. It supports high-precision dynamic identification, detection, and correction of errors during high-speed movement, thereby enhancing overall system performance. The experimental results demonstrate exceptional accuracy, particularly in detecting small parts measuring 28.87 mm in length and 12.36 mm in width, achieving an impressive precision of 0.04 mm. Additionally, our system boasts meticulous hardware selection, robust software stability, and high-performance capabilities, culminating in improved detection efficiency and accuracy. This research not only contributes novel ideas and results but also holds significant commercial value in industrial applications. Overall, our proposed methodology represents a noteworthy advancement in automatic inspection systems, offering superior performance and reliability compared to existing approaches.