Intelligent monitoring and fuzzy control of MIG welding seam tracking based on passive visual sensing

Abstract

To reduce the risk of personnel operation and further improve welding efficiency, weld seam tracking in MIG welding process arc has to be developed for automatic manufacturing. Weld seam tracking system mainly contains intelligent monitoring and fuzzy control. For monitoring part, an optical testing platform and a passive visual detecting device are established to analyze groove and arc position. Also, preprocessing workflow and adaptive enhancement algorithm are built to increase image gray values. Deep learning program is used to select and locate interest area to improve the accuracy of detection. The arc position calculation model is also proposed to extract geographic location. For control part, based on welder’s operation skills, fuzzy logic rules are programmed to control the arc position at the middle of gap. Also, control experiments are carried out and compared with manual adjustment. Results show that: (1) with preprocessing workflow and adaptive enhancement algorithm, the average gray value of the groove area and the arc area increased by 114% and 100%; (2) by using deep learning, the interest area contains information of groove shape and oscillating arc position and could be selected accurately, and the mAP index is as high as 99.27%; and (3) based on the preset deviation test, the pixel error of the alignment deviation detection is within 8 pixels. And with the alignment deviation, distance can be controlled between ± 0.5 mm.