Computer Vision System for Welding Inspection of Liquefied Petroleum Gas Pressure Vessels Based on Combined Digital Image Processing and Deep Learning Techniques.

摘要: One of the most important operations during manufacturing process a pressure vessel is welding. The result this operation has great impact on integrity; thus, welding inspection procedures must detect defects that could lead to an accident. This paper introduces computer vision system based structured light for liquefied petroleum gas (LPG) vessels by using combined digital image processing and deep learning techniques. procedure applied prior was convolutional neural network (CNN), it correctly detected misalignment parts be welded in 97.7% cases method testing. post-welding laser triangulation method, estimated weld bead height width, with average relative errors 2.7% 3.4%, respectively, allows us geometrical nonconformities compromise integrity. By system, quality index improved from 95.0% 99.5% practical validation industrial environment, demonstrating its robustness.