Xiaolei Fang
North Carolina State University, NC·Engineering
Factories rely on data to catch problems before machines fail or parts come out defective. But one plant rarely has enough records of faults and failures to train good prediction models, and companies are reluctant to share confidential data.
- Fields
Engineering
Mathematics & Statistics
Computer Science
- Topics
federated learning
privacy-preserving machine learning
statistical learning
- Grants
- 2 known grants, through Jul 2027
What they work on
- Develop a federated learning framework that lets distributed manufacturing plants jointly train condition-monitoring models while keeping each plant's data local, using robust low-rank learning for data curation and supervised dimension reduction for shared features.
- Create classification-based anomaly detection, regularization-based fault diagnosis and parametric prognostic models that work on time series, profile and image-stream data from industrial equipment.
- Design federated optimization algorithms that support operational and maintenance decisions across geographically separated manufacturing systems.
- Build a computationally efficient statistical learning model that detects defects in real time during electron beam powder bed fusion (EB-PBF) metal printing from electron-based sensing data.
- Develop a privacy-preserving federated, regularization-based model that pinpoints which EB-PBF process parameters cause defects, and validate it with designed printing experiments.
More about the research
This research develops statistical and machine learning methods that let manufacturers learn together while keeping their data private. It also builds real-time quality control for metal 3D printing.
Why it matters
Manufacturers could catch faults earlier, predict equipment failures and cut costs by pooling knowledge across plants without exposing confidential data. For metal additive manufacturing, real-time defect detection and root-cause diagnosis could reduce defective parts. This could help EB-PBF printing be adopted more widely for producing metal components.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
A Privacy-Preserving Collaborative Condition Monitoring and Decision-Making Methodology for Distributed Manufacturing Systems
NSF · Aug 2024 – Jul 2027
FMSG: Cyber: Process Monitoring Methods for the Product Quality Improvement of Electron Beam Powder Bed Fusion Additive Manufacturing Processes
NSF · Oct 2022 – Dec 2026
Closest labs
Yiliang Liao
Iowa State University, IA·Engineering
Determine how the evolving, uneven thermal environment of a hot liquid support bath controls polymer diffusion and interlayer fusion in embedded printing, using deep-neural-network molecular dynamics, phase-field modeling, and experiments.
Albert To
University of Pittsburgh, PA·Engineering
Build a framework that couples finite element heat simulation with data-driven surrogate models through a local thermal environment representation, allowing two-way information exchange to predict melt pool evolution quickly in laser powder bed fusion.
Lili Su
Northeastern University, MA·Computer Science·Early-career
Build theory for federated learning that uses the statistical structure of each device's data, to explain why it succeeds in practice where optimization-only analyses predict failure.
The school
North Carolina State University: 40 labs on PI Pages, $717M a year on research, 603 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.