G. Melton, D. Bertaso, TWI Ltd, Cambridge, United Kingdom

Summary:

In mechanised and robotic GTA welding of aerospace components, variations in weld bead geometry may occur due to many factors including variable fit-up, heat build up, variations in heat sinking e.g due to changes in cross section and jigging, electrode erosion and cast to cast variations. In a recently completed EU funded project (MMFSC) a vision based control system has been developed which overcomes these variations to improve repeatability and provide consistent weld bead geometry.

The control system uses a computer vision technique based on the real-time measurement and control of the upper surface or ‘topface’ of the weld pool. A novel reference feature correlation based image analysis algorithm has been developed that may be configured to operate with a number of different equipment and welding processes. Upper surface weld pool width regulation via a simple classical feedback control algorithm has then been demonstrated to show how this approach may be used to improve weld uniformity and repeatability.

The control system has been validated on Inconel alloy 718 over the thickness range 1 mm to 3.2 mm as is typically used in aeroengine manufacture. Following initial trials at the TWI laboratories in the UK, the system was interfaced to an ABB robot and ServRobot laser scanner for end user trials in Sweden.

This paper further describes the vision and computer controlled system and the validation of the system on a range of Inconel 718 test components.