Kalman Filtering Compensated by Radial Basis Function Neural Network for Seam Tracking of Laser Welding

An approach for seam tracking during high-power fiber laser butt-joint welding is presented. Kalman filtering (KF) improved by the radial basis function neural network (RBFNN) of the molten pool images from a high-speed infrared camera is applied to recursively compute the solution to the weld position equations, which are formulated based on an optimal state estimation of the weld parameters in the presence of colored noises. This NN could suppress the filter divergence and improve the system robustness. In comparison with the traditional KF algorithm, the actual welding experiments demonstrate that the KF compensated by RBFNN is more effective in improving the seam tracking accuracy and lessening the disturbance influences caused by colored noises.