CyPLAM, Cyber-Physical Laser Metal Deposition
Existing image-based monitoring and control approaches in laser processing resort to embedded and PC-based platforms. Flows of data easily achieve tens of MB per second, requiring a strong selectivity in an early stage. Few features are gathered to control in RT usually a single parameter (e.g. laser power) or to be used in a monitoring system for quality control. However, gathering and processing large temporal series of data in some detail becomes prohibitive for this kind of systems. As regards to additive manufacturing by Laser Metal Deposition (LMD), this is of great interest since it is a long process (e.g. lasting hours) that may accumulate important thermal and dimensional deviations. This results in the need of reworking after the cladding process, which means waste of material, time and energy. Moreover, it makes difficult to ensure internal mechanical properties of produced parts. A good example of these challenges may be found in repairing of stamping molds for the automotive sector.
Processing data (thermal high-speed image sequences and 3D profiles) in the cloud will bring the opportunity to gather large amounts of data and to use machine learning techniques to extract information and to learn interrelations between relevant process parameters. CyPLAM will rely on this approach to:
-
Adjust parameters during the process, so that deviations in a given track can be corrected in the next layer.
-
Quality diagnosis and process reconfiguration from large series of data from manufacturing of previous parts.
Acknowledgement
This work is been supported by the European Commission through the research project "Factories of the Future Resources, Technology, Infrastructure and Services for Simulation and Modelling 2 (FORTISSIMO 2)", H2020 - Grant Agreement Nº 680481.