No.1018E 2022Special Edition on Industry Related Technologies Automation of Magnetic Particle Inspection Process

Category TECHNICAL REPORT
Author Y. ARAI
Production Engineering Administration Dept., Production Engineering Division
K. FUSAYASU M. SUGIHARA
Process innovation Engineering Dept., Production Engineering Division
Abstract In automating the magnetic particle inspection process for rack and pinion type steering racks, the problems were the false indication generated on the ridgeline of rack teeth and the variety of defect modes. Focusing on the strength of magnetization and fl ow of magnetic particles, we have developed a magnetic particle application technology suitable for image recognition. In addition, by using deep learning for image recognition, we were able to overcome these challenges. By combining magnetic particle application technology and image recognition technology, a recognition accuracy of 0% nondetection and less than 5% over-detection is achieved.
Keyword magnetic particle testing, automation, deep learning, image recognition
Search tag

Download Ranking of
ENGINEERING JOURNAL