Combining Physical Simulation and Bayesian Optimization for LMD Process Parameter Development

Produktinformationen

Dr.-Ing. Thomas Schopphofen, thomas.schopphoven@ilt.fraunhofer.de; M.Sc. Max Zimmermann, max.zimmermann@ilt.fraunhofer.de; Dr. rer. nat. Norbert Pirch; M.Sc. Viktor Glushych,viktor.glushych@ilt.fraunhofer.de;
Fraunhofer Institute for Laser Technology ILT; 

https://doi.org/10.53192/ITSC2026526

Laser Material Deposition (LMD) is an advanced manufacturing technology that en-ables the deposition of metallic coatings at high productivity and with high precision. Compared to conventional methods, LMD offers significant advantages in terms of efficiency, coating quality, and material savings. However, realizing these benefits requires meticulous control and optimization of numerous interdependent process parameters, many of which are nonlinearly coupled. The lack of a reliable, predictive process model further complicates model-based optimization approaches. To ad-dress these challenges, Bayesian Optimization (BO) is applied to systematically tune LMD process parameters. BO leverages probabilistic models, trained on experimental data, to efficiently explore the parameter space and predict optimal settings based on past outcomes. To improve efficiency and reduce the number of required experiments, physical simulations are used in advance to narrow down the feasible parameter space, thereby guiding the optimization process toward promising re-gions. This study explores the application of Bayesian Optimization for LMD process tuning and demonstrates its potential to enhance both the stability and quality of the coating process. Preliminary results indicate that BO, in combination with physical simulation, is a promising strategy for improving process robustness and perfor-mance in LMD.

Reihe

PDF-Download zu https://doi.org/10.53192/ITSC2026526

Erscheinungsdatum

März 2026

Bindung

PDF Download

Seiten

6