Abstract: The manual welding of tunnel steel segment linings is characterized by uncontrollable welding quality and low efficiency. To address these issues, three-dimensional(3D) machine vision technology is employed to identify weld joints in steel segment linings, enhancing traditional automatic welding technology through compensation and deviation correction. The automatic welding process is further refined by incorporating digital simulation of weld characteristics, joint identification, coding, and welding techniques. The findings indicate the following: (1) Assisted by 3D visual sensing technology, automatic welding achieves precise identification of weld joints in steel segment linings. In comparison to manual welding, joints welded automatically are continuous, neat, and uniform, exhibiting no micro-cracks or micropores. Additionally, the cross top angle of the weld joint at the bottom corner of the steel chamber structure is smooth and neat, with no leakage observed after water pressure testing at 0.6 MPa. (2) During the current testing phase, the automatic welding efficiency is 45 h per block simplex. Following welding process optimization and manipulator adjustments, this efficiency is projected to improve to 20 h per block simplex.

Key words: metro tunnel, steel segment lining, automatic welding, three-dimensional machine vision