Monitoring and Control of Robotized Laser Metal-Wire Deposition
Doctoral thesis, 2012
The thesis gives a number of solutions towards full automation of the promising manufacturing technology Robotized Laser Metal-wire Deposition (RLMwD). RLMwD offers great cost and weight saving potentials in the manufacturing industry. By metal deposition is here meant a layered manufacturing technique that builds fully-dense structures by melting metal wire into solidifying beads, which are deposited side by side and layer upon layer. A major challenge for this technique to be industrially implemented is to ensure process stability and repeatability. The deposition process has shown to be extremely sensitive to the wire position and orientation relative to the melt pool and the deposition direction. Careful tuning of the deposition tool and process parameters are therefore important in order to obtain a stable process and defect-free deposits.
Due to its recent development, the technique is still manually controlled in industry, and hence the quality of the produced parts relies mainly on the skills of the operator. The scientific challenge is therefore to develop the wire based deposition process to a level where material integrity and good geometrical fit can be guaranteed in an automated and repeatable fashion.
This thesis presents the development of a system for on-line monitoring and control of the deposition process. A complete deposition cell consisting of an industrial robot arm, a novel deposition tool, a data acquisition system, and an operator interface has been developed within the scope of this work. A system for visual feedback from the melt pool allows an operator to control the process from outside the welding room. A novel approach for automatic deposition of the process based on Iterative Learning Control is implemented. The controller has been evaluated through deposition experiments, resembling real industrial applications. The results show that the automatic controller increases the stability of the deposition process and also outperforms a manual operator.
The results obtained in this work give novel solutions to the important puzzle towards full automation of the RLMwD process, and full exploitation of its potentials.
Additive Layer Manufacturing
Iterative Learning Control
Robotic Weld Equipment
Laser Metal Wire Deposition