Seam tracking is significant to obtain good welding quality. Aiming at establishing a model to detect and calculate the seam tracking offset during high-power fiber laser welding of Type 304 austenitic stainless steel plate butt joint welding, an infrared sensitive high-speed camera arranged off-axis orientation of laser beam was applied to capture the dynamic thermal images of molten pool. Six parameters such as the keyhole configuration parameters (include four parameters), keyhole centroid parameter and heat accumulation parameter were defined as the eigenvalues of seam tracking offset to determine the seam tracking offset between the laser beam and the desired welding trajectory. A BP neural network model was built to reflect the correlations between the defined eigenvalues and the seam tracking offset. The welding experiments confirmed that the seam tracking offset between the laser beam focus and the welding seam could be monitored and calculated by the BP neural network model effectively.
Back-Stepping and Neural Network Control of a Mobile Robot for Curved Weld Seam Tracking
