"Transfer learning for nondestructive fatigue life prediction of additively manufactured metals in a distributed manufacturing paradigm"

Abstract: Understanding the fatigue behavior and accurately predicting the fatigue life of laser powder bed fusion (L-PBF) parts remain a pressing challenge due to complex failure mechanisms, time-consuming tests, and limited fatigue data. I will first introduce a physics-informed transfer learning framework to understand process-defect-fatigue relationships in L-PBF by integrating process parameters, XCT-inspected defects, and fatigue test conditions. It aims to leverage a pre-trained model with abundant process and defect data from the source task to predict fatigue life non-destructively using limited fatigue test data from the target task.

Moreover, to facilitate privacy-conserving data sharing and modeling among distributed manufacturers, we proposed a personalized federated transfer learning framework (FedCOT). FedCOT enables "target" organizations with limited features to benefit from "source" organizations with sufficient features in terms of prediction performance, while maintaining data privacy through a central server. Each organization learns a personalized encoder-regressor structure. Target organizations align their latent representations' structure and corresponding responses with the source organizations' information. FedCOT achieves a latent space where target organizations are highly aligned with source organizations, resulting in a significant 34.99% improvement in prediction performance over the baseline method.

Dr. Jia “Peter” Liu

Associate Professor, Trey Lauderdale Industrial and Systems Engineering Fellow

University of Florida

Speaker Bio: Jia “Peter” Liu is an associate professor and the Trey Lauderdale Industrial and Systems Engineering fellow at the University of Florida. His research interests encompass statistical learning and machine learning with applications in advanced manufacturing. He works to integrate physics knowledge and interpretable data-driven modeling for complex manufacturing processes involving heterogeneous sensors, with a primary focus on understanding the fatigue performance of powder bed fusion, particularly in the aerospace, defense, and automotive sectors. His research has been funded by NSF, DoD, FAA, and NIST, and he has been honored with several awards, including the 2024 ASME Rising Star of Mechanical Engineering and the 2023 NSF CAREER Award. He is a senior member of INFORMS and a member of IISE, ASME, and SME. He holds a Ph.D. in Industrial and Systems Engineering, an M.S. in Statistics from Virginia Tech, and a B.S. and M.S. in Electrical Engineering from Zhejiang University, China.