Abstract:

In order to realize the real-time and on-line state monitoring of the high-frequency electric resistance welding ( HF-ERW) ,according to the analytical results of various welding phenomena as well as the welding defects and their causes,an automatic monitoring method based on the high-speed CCD imaging and the digital image processing is proposed. In this method,digital images of the welding region are obtained by using a CCD camera set on a fixed location above the welding region,and the parameters,such as the characteristic heating area,the value of V-type angle,the location of welding bond point and the symmetry degree of the heating region,are measured and processed through a multi-characteristic fusion,thus revealing the current welding state. Moreover,during the image processing,a new segmentation algorithm based on R-G-B non-equilibrium anti-interference technology is employed,the V-type angle detection and the corresponding bisection algorithm based on line fitting are adopted,a function to evaluate the symmetry degree of the heating region is designed,and a multi-characteristic fusion model based on BP neural network is established. Field test results show that the proposed method helps to effectively monitor the state of HF-ERW and control the welding quality.

Key words: high-frequency resistance welding, image processing, filling, V-type angle, symmetry degree