Monitoring of butt weld penetration based on infrared sensing and improved histograms of oriented gradients

In the process of butt welding, achieving real-time monitoring of butt weld penetration is of great significance for improving forming quality. Currently, weld penetration detection methods based on weld pool vision are easily disturbed by arc lights. In this study, an innovative approach for butt weld penetration detection using infrared sensing and improved histograms of oriented gradients (HOG) is proposed. First, an improved HOG algorithm was studied to extract the temperature field distribution features of the welding heat-affected zone (HAZ) around the weld pool. Second, the feasibility of using the improved HOG feature to detect weld penetration was analysed, and a weld penetration detection model was established. Finally, the generalisation performance of the proposed weld penetration detection algorithm was confirmed, and the influence of the location and size of the detection area on the accuracy of the detection algorithm was analysed. The testing results indicate that the proposed detection algorithm exhibits high precision and excellent generalisation performance, and enhances the means of monitoring of butt weld penetration.