A New Optical Measurement System for Noncontact Crack Detection in Metal Sheets Shaped by Deep Drawing Process

Abstract

In the automotive industry, sheet metal products are generally shaped by deep drawing process. Especially after the deep drawing process of stainless-steel sheet metal products, fine splits or surface cracks occur in the tensile and compressive strain regions. The reason for this situation can be given as the errors in the mold, the mechanical properties of the sheet metal varying according to the producing company, and the difference of this property in different parts of the sheet roll. If a deformed and pre-inspected sheet metal is mounted on an automobile, painted, and then put into a drying oven, all defects in the sheet metal will become evident as it expands under high temperature. Detecting defects at this stage increases the cost. Accordingly, in this study, a new real-time optical inspection system based on multiple photosensitive resistors (LDRs) was developed for crack detection in the automotive industry. In this system, a light source is placed on the inside of the crankcase cover in a dark, light-proof environment. Then, LDR sensors are placed around the outside of the cover to detect the level of light leaking out of the cover. At the same time, the environment was monitored with a camera, and image processing was performed with OpenCV. The collected data was sent to a cloud database server in JSON format over Wi-Fi with NodeMCU and recorded. Finally, the results of the experiment were analyzed with a WebGL-based web interface developed using Unity and visualized on a crankcase cover model drawn with Solid Works. The researcher can visually see the manufacturing defects and dimensions in the 3D environment on the model. It was determined that the system can identify the crack in a crankcase cover shaped by the deep drawing process and verify it with the camera within 6 s.