Multi-Laser Scan Assignment and Scheduling Optimization for Large Scale Metal Additive Manufacturing

Abstract

Metal additive manufacturing (AM) has attracted significant attention in various industry sectors for large-scale fabrication. However, the limited fabrication efficiency has hindered its practical implementation. In comparison to traditional methods of tuning process parameters, concurrent AM equipped with multiple independently driven lasers is a more promising technique recently developed for the efficient fabrication of large metal parts. To maximize fabrication efficiency while ensuring quality for multi-laser AM processes, an optimization problem is proposed in this work for multi-laser scanning plan, including scan vector assignment and scheduling. The goal is to minimize the makespan while considering factors that may affect the quality of metal AM parts as constraints. Specifically, the constraints associated with heat-affected zones (HAZs) and the user-specified single-laser scanning area are considered. The optimization model is solved by deep reinforcement learning (DRL), offering the flexibility to include or exclude considerations for different quality/process requirements. Two case studies demonstrate the application of DRL models considering different sets of constraints and compare their performance with two baseline scheduling methods in terms of fabrication efficiency and violation of quality constraints. In addition, the impact of the laser number on operational improvement and the computational cost of the DRL model is also studied.

Keywords:

• Metal additive manufacturing
• multi-laser
• scan assignment and scheduling
• deep reinforcement learning

Disclaimer

As a service to authors and researchers we are providing this version of an accepted manuscript (AM). Copyediting, typesetting, and review of the resulting proofs will be undertaken on this manuscript before final publication of the Version of Record (VoR). During production and pre-press, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal relate to these versions also.

Additional information

Notes on contributors

Yuxin Yang

Yuxin Yang received her BS degree in Safety Engineering from the China University of Labor Relations in Beijing, China, in 2019. She earned her master’s degree in Industrial Systems Engineering from Binghamton University in 2021. She is currently pursuing her Ph.D. in the Department of Systems Science and Industrial Engineering at Binghamton University. Her research focuses on reinforcement learning, data-driven analytics, optimization and simulation, and advanced manufacturing.

Lijing Yang

Lijing Yang received his bachelor's and master's degrees in Electrical and Computer Engineering, with emphases in signal processing, machine learning, and deep learning, from the Ming Hsieh Department of Electrical and Computer Engineering at the University of Southern California, Los Angeles, CA, in 2022 and 2023, respectively. His research interests include DSP, Deep Reinforcement Learning, and Sound Recognition & Generation.

Abdelrahman Farrag

Abdelrahman Farrag received his master’s degree from the Department of Mechanical Design and Production, Faculty of Engineering, Assiut University, Egypt, in 2019. He is currently pursuing his Ph.D. degree in the Department of Systems Science and Industrial Engineering, Binghamton University, Binghamton, NY, USA. His research interests include data-driven and physics-informed predictive modeling in smart and advanced manufacturing, modeling and simulation, and optimization.

Fuda Ning

Dr. Fuda Ning is an Assistant Professor in the Department of Systems Science and Industrial Engineering at Binghamton University. His research interest lies in additive manufacturing processes with a technical focus on the understanding of multi-scale interactions between energy input and matter through computational, analytical, and experimental efforts. His research has been sponsored by NSF, SEMI-FlexTech, and industry with a total funding of around $2.8 million. Ning’s research has led to over 90 academic publications with more than 5,600 citations and an H-index of 36. Ning owns 11 journal article publications with an impact factor higher than 10. Due to his impactful research, he was among Stanford’s Top 2% Scientist List in the world since 2021. He is also an editorial board member for five journals in the field of manufacturing.

Yu Jin

Dr. Yu “Chelsea” Jin is an Assistant Professor in the Department of Industrial and Systems Engineering at the University at Buffalo. Her research focuses on cutting-edge techniques and methodologies, especially machine learning and artificial intelligence methods, to enhance advanced manufacturing systems within the context of Industry 4.0. Her research has been sponsored by the Transdisciplinary Area of Excellence Seed Grant, Integrated Electronics Engineering Center, and industry. The research findings of her group have been published in IISE Transactions, ASME Journal of Manufacturing Science and Engineering, Rapid Prototyping Journal, IEEE Transactions, etc. She has been an active member of IISE, INFORMS, ASME and Alpha Phi Mu. She has been a board member of IISE DAIS Division from May 2022 – May 2024 and an officer of INFORMS QSR Fund Raising Committee since May 2023.