Seam Tracking Based On Magneto-Optical Imaging And Adaptive Kalman Filtering Algorithm

In laser butt welding process, precise control of laser beam welding is always on the premise of laser welding quality assurance. Since the laser beam spot diameter is small (typically less than200μm), sensitive to the size of the weld gap, the weld can be as small as required. The traditional method of structured light vision, the structure will be visible spectral bands of light across the weld location, thus forming mutations characteristic structure of the light in the weld and the use of image processing techniques to accurately identify the weld location. However, due to the traditional method of structured light vision in the weld gap is less than0.1mm when the weld will not effectively track, resulting in a precise control of micro-gap welding and laser beam aligned with the weld path is very difficult to track.This topic extracted through magneto magneto-optical image sensor weld, the weld characteristics in magneto-optical image study accurately detect the weld seam tracking center position and achieve. Faraday effect in the theoretical basis of the magneto-optical imaging technology, is a small surface defects can be visualized detection approach. By analyzing the parameters of the magneto-optical image of gray under the weld, using the particle filter and mean shift algorithm to track the actual location of the weld testing. Experimental results show that the two methods can effectively extract the weld location.Secondly, a lot of noise in the welding process due to changes in temperature welding material, the frequency of the alternating magnetic field magneto-optical imaging sensor excitation coil, the coil spacing and weldments arising from such factors seriously affect the accuracy of seam tracking. In this regard, this paper studies the system state Kalman filter based optimal estimation method. Weld path prediction model by establishing real-time measurements with a laser beam deviation and tracking, eliminating noise interference and improve seam tracking accuracy. Laser welding process test carbon stainless steel as test subjects to study the use of Kalman filtering in colored noise environments for optimal weld center position estimation, the best predictive value of the weld center position, eliminate noise and measurement system status interference. Magneto-optical images extracted from the weld location parameters and constitute state vector, create a system state equation and measurement equation weld seam position based on the location parameters. This paper studies the system processes are noise and color noise measurement noise when using Sage adaptive filtering, which uses a new interest before m-step sequence estimation process noise covariance matrix of the traditional Kalman filter algorithm is improved. Experimental results show that the improved Kalman filter algorithm can effectively suppress the color noise and seam tracking can improve accuracy.