Method of Process Optimization for LMD-Processes using Machine Learning Algorithms

• Published in: Procedia computer science Vol. 217; pp. 1506 - 1512
• Main Authors: Gröning, Holger, Zenisek, Jan, Wild, Norbert, Huskic, Aziz, Affenzeller, Michael
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 2023
• Subjects: Additive Manufacturing; Domain Knowledge Integration; Laser Metal Deposition; Machine Learning; Process Improvement; Laser Metal Deposition; Process Improvement; Additive Manufacturing; Machine Learning; Domain Knowledge Integration
• ISSN: 1877-0509; 1877-0509
• DOI: 10.1016/j.procs.2022.12.350

Additive manufacturing processes such as laser metal deposition have great potential for supplementing or substituting established manufacturing processes. One of the challenges is the time-consuming development of thermally stable process conditions for the production of defect-free components. The presented approach aims for extracting all relevant machine-, process- and result parameters from different sources, using an algorithm to merge the heterogeneous data streams and to stabilize and optimize the process through machine learning algorithms. To provide qualified training data an expert system is presented to facilitate the identification of causal relationships by domain experts. Furthermore, the strategy focusses on the development and implementation of virtual sensors to avoid the hardware- and software-integration effort, to reduce the amount of data processed and enable for an in-situ closed-loop optimization.