Recognition Of Characteristics Of Plume And Spatter Images During High-power Disk Laser Welding

Laser welding is an important processing mode of welding parts during21century. The laser energy is highly concentrated, metal can cooling fast after the metal part been heating rapidly and then been processed. The thermal deformation is so small compared with many other parts, and the thermal effect is small too. It can be used to weld the refractory metal and metal that difficult to weld, such as titanium alloy, aluminum alloy, etc. In addition, laser welding process has no pollution to the environment and the technology is easy to master. Thus, laser welding technology has a more and more widely use in the field of electronics, national defense, instrumentation, batteries, medical equipment and many other industry. High power disk laser welding is one of the most advanced laser welding technology at present, which has advantages such as high power density, good beam quality, big depth and width ratio and high utilization rate of laser.Metal vapor plume and spatter are important phenomenon which produced during the process of high power disk laser welding. They have a close relationship with the welding quality and stability. So, there is a significant importance of the research on the characteristics recognition technology of metal vapor plume and spatter images, it is an important foundation to realize the real time control of welding process. Stainless steel304was taken as the test object of high power disk laser welding. A high-speed camera was used to capture the ultraviolet band and visible light band plume and spatter images during the laser welding. Image processing techniques such as median filtering, Wiener filtering, lightness transform, graylevel threshold and edge detection were applied to process those images. The tangent value of the angle between the straight line that connect the centroid of metal vapor plume and the welding point and the horizontal axis, the abscissa of the highest point of plume, the density of the plume, the polar coordinates of the centroid of metal vapor plume(radius vector and polar angle), the area and numbers of spatters, the average gray level of a spatter grayscale image and the entropy of a spatter grayscale image those9characteristic parameters of metal vapor plume and spatter were defined and extracted to study the relationship between the fluctuations of those characteristic parameters and welding quality. In order to establish a recognition model between the plume and spatter characteristic parameters and welding quality, each choose7different characteristic parameters to form a7-demonsional eigenvector, then principal component analysis method and K-L transform method was applied respectively to transform the7characteristics into a new set of characteristics, the actual weld width was taken as the measurement factor of the welding quality and welding stability and the metal vapor plume and spatter images were classified into two types accordingly, k nearest neighborhood method was used to recognize the metal vapor plume and spatter image automatically. Experimental results showed that the change of the characteristic parameters of metal vapor plume and spatter were consistent with the actual welding process, using k-nearest neighbor method could effectively recognize the two types of images based on the results of principal component analysis and K-L transform. This will make contribution to the monitoring of laser welding process in real-time.