Research On Welding Pool Characteristics Of Short Circuit Welding Based On Digital Image

With the progress in science and technology, welding automation has become an irreversible developmental direction of modern welding technology. And controlling of the welding process is an important component of welding automation. As we know, sensing of welding information is the premise of welding process control. So only by getting accurate information during the welding process and monitor characteristic parameters of the welding process on-line, we can control the welding process effectively. Study on welding pool image of short circuit welding is the basis of improving welding technics, researching short circuit transition principle and achieving intelligent control.Welding pool of short circuit welding is dynamic with high speed and has a lot of splash. Meanwhile the strong arc results in a very large dynamic range of light intensity around the welding pool. All of these make it very difficult to get a clear image of the welding pool. A passive visual sensing system of welding pool is originally designed and built by using a special CMOS camera with LinLog technology. We try to take photos of the welding pool using this system from the rear with the reflected arc and its own radiation of light as a light source. LinLog technology can enlarge the dynamic range of the camera and compress the strong arc. Under the mode of global shutter and high-speed photography, welding pool images are obtained with this system. Edge of the welding pool and detail inside the welding pool in the images are clear, which makes it possible to do further analysis of the welding pool image of short-circuit welding.According to the character of the images, a set of image processing method is originally designed. After image pre-processing (threshold, particle filtering and morphological closing operation) and image post-processing (edge detection and polynomial edge fitting), images with complete edge and without arc are gotten. On this basis, several parameters are presented and extracted to describe the features of the welding pool and change of the welding pool characteristics under different welding condition. The parameters, such as area and ellipticity of the welding pool, average gray value and surface roughness of the welding pool, oscillation amplitude of the welding pool, characterize the welding pool from different aspects respectively, including geometric shape, gray level distribution, and change of position of the special point of the welding pool.Finally, the system is applied on short circuit welding of low-carbon steel under different welding conditions. The acquired images are processed with the aforementioned image processing methods. After that, parameters of welding pool are extracted and relationship between characteristics of welding pool and welding conditions is explored. Experiments show that in welding process, the change of heat input can cause change of welding pool size and shape; gray distribution of welding pool and droplet transition have the same period and frequency; when the welding wire offsets from the gap of workpieces, the welding pool shows obvious features. Since this kind of study is still on the exploratory stage, there is much improvement and research work to do for the reliability and accuracy of the sensing of welding pool characteristics.