Abstract: Aiming at the problem of system noise and measurement noise existing in the butt-welding process of the V-groove plate, the slope analysis was combined with the Kalman filter to weld tracking the seam. A laser vision sensing system was used to collect a stripe image of a V-groove. By obtaining the internal and external parameters of the system, a coordinate transformation model for weld feature points was established by the sawtooth target. And a boundary-weighted squared weighted center of gravity method based on the extremum method was proposed to extract the stripe centerline. The slope analysis method was used to locate the inflection point of the center line of the laser stripe. The center line of each stripe was fitted by the least squares method to obtain the weld feature points. Then the influence of measurement noise was processed by Kalman filter. The butt-welding test of the V-groove plate was carried out and the torch could always be aligned with the center of the weld. The results indicate that the seam tracking method based on slope analysis and Kalman filter can reduce the influence of system noise interference, achieve accurate extraction of weld feature points, and improve weld seam tracking accuracy.
Key words: seam tracking; weld feature points; slope analysis method; Kalman filter