Weld deviation prediction model based on time series and Kalman filter

The poor stability of the arc in narrow gap welding restricts the application of arc sensors in seam tracking. To address this issue, a prediction model based on time series analysis and Kalman filter was established to enhance the anti-interference ability and improve the tracking control accuracy of arc sensors in narrow gap welding. By analyzing the causes of arc instability, sensing data fluctuated greatly when the biting edge phenomenon and arc breakage occurred, which is detrimental to weld tracking. To eliminate abnormal sensing data, a corresponding time series prediction model was constructed to capture the change rule of sensing data values over the swing cycle. Furthermore, the Kalman filter was employed to reduce interference and noise to address the issue that sensing data values were easily affected by many unstable factors during the sampling process. By combining time series analysis and Kalman filtering, a hybrid prediction model was developed to solve the problem of unstable sensing data caused by arc instability. The results have shown that the hybrid prediction model can effectively predict and correct unstable data, improve the accuracy of weld tracking, and provide a theoretical foundation for weld prediction and tracking.