Torch-mounted sensor detects weld defects in real time

Sonibel Instruments has introduced a real-time weld defect detection system designed to help fabricators identify and correct issues during the welding process. The compact sensor mounts directly onto the welding torch and uses high-fidelity sound analysis to detect defects as they occur, enabling immediate in-process repair.

By allowing welders to correct flaws before inspection, the technology helps reduce costly rework, minimize project delays, and improve first-pass inspection rates.

The system uses machine learning to distinguish between various weld defects in noisy industrial environments, capturing meaningful acoustic signals amid background interference. Detected issues are displayed instantly on a screen mounted to the welding unit, allowing welders to take corrective action on the spot rather than waiting for post-weld testing.

Current optimization supports detection of porosity and lack of fusion in semi-automatic MIG welding applications. The technology can also be customized for other processes and defect types, including TIG, flux core, sub arc, stick, and laser welding, across both semi-automatic and automated operations.