Dual-loop Adaptive Iterative Learning Control for Robust Temperature Stabilization in Selective Laser Melting
Event details
| Date | 20.02.2025 |
| Hour | 11:00 › 12:00 |
| Speaker | Mr. Junan Lin, Master Student @ ETH Zurich |
| Location | |
| Category | Conferences - Seminars |
| Event Language | English |
Abstract:
Selective laser melting (SLM) is a metal additive manufacturing process. Temperature stabilization in SLM is of vital importance to avoid defects such as distortion and cracking. Existing control methods require manual tuning under potential model mismatch, and they can fail when printing complex geometries. This paper introduces a dual-loop control strategy to stabilize the surface temperature for robustness and near-optimal performance in the presence of model mismatch and measurement noise. The proposed method integrates in-layer feedback control via linear output feedback with gains pre-optimized by a policy gradient method, and layer-to-layer feedback control that combines temperature trajectory optimization and iterative learning control. Simulation results show that the dual-loop controller stabilizes the temperature well for each surface layer, even under significant model mismatch and measurement noise. Experiments on a physical SLM platform demonstrates that the proposed simulation-based auto-tuning approach achieves performance on par with a state-of-the-art in-situ-tuned controller.
Bio:
Mr. Junan Lin is a master student in Robotics, Systems and Control program at ETH Zürich, and is expected to graduate in May, 2025. He received his bachelor degree in Automation in 2022 from Beihang University, China. His research interests include control theory, machine learning and their applications. Currently, he is doing his master thesis about approximating optimal control in infinite horizon constrained LQR by short-horizon MPCs, in Automatic Control Laboratory (IfA) at ETH Zürich.
Selective laser melting (SLM) is a metal additive manufacturing process. Temperature stabilization in SLM is of vital importance to avoid defects such as distortion and cracking. Existing control methods require manual tuning under potential model mismatch, and they can fail when printing complex geometries. This paper introduces a dual-loop control strategy to stabilize the surface temperature for robustness and near-optimal performance in the presence of model mismatch and measurement noise. The proposed method integrates in-layer feedback control via linear output feedback with gains pre-optimized by a policy gradient method, and layer-to-layer feedback control that combines temperature trajectory optimization and iterative learning control. Simulation results show that the dual-loop controller stabilizes the temperature well for each surface layer, even under significant model mismatch and measurement noise. Experiments on a physical SLM platform demonstrates that the proposed simulation-based auto-tuning approach achieves performance on par with a state-of-the-art in-situ-tuned controller.
Bio:
Mr. Junan Lin is a master student in Robotics, Systems and Control program at ETH Zürich, and is expected to graduate in May, 2025. He received his bachelor degree in Automation in 2022 from Beihang University, China. His research interests include control theory, machine learning and their applications. Currently, he is doing his master thesis about approximating optimal control in infinite horizon constrained LQR by short-horizon MPCs, in Automatic Control Laboratory (IfA) at ETH Zürich.
Practical information
- General public
- Free
Organizer
- Professor Giancarlo Ferrari Trecate
Contact
- barbara.schenkel@epfl.ch