Dynamic synchronization and feature region fusion for real time girth weld tracking

Narrow-gap all-position girth welding of pipelines requires accurate seam tracking to handle spatial complexity and dynamic arc behavior. Since existing methods struggle with signal boundary extraction and real-time control, this study introduces a weld-tracking system that integrates torch oscillation synchronization with current feature-region analysis. Pulse (P) and direction (D) signals were used to dynamically partition the welding current into three regions (left edge, middle, and right edge), which resolved boundary uncertainties. Vertical tracking used a binary linear regression model that correlated the wire feed speed and contact tip-to-work distance (CTWD) with middle-region current. Horizontal tracking used trapezoidal integration to quantify current differences between edge regions. The experimental results showed a steady-state accuracy of 97% and a dynamic response time under 50 ms, achieving a precision of ± 0.10 mm. Cosine-curve displacement control enabled overshoot-free adjustments with steady-state errors below 2.2%. This work enhances automated welding systems by combining dynamic signal synchronization with algorithmic integration, offering a new framework for real-time control in narrow-gap welding applications.