Skip to main content
  • Paul Scherrer Institut PSI
  • PSI Research, Labs & User Services

Digital User Office

  • Digital User Office
  • DE
  • EN
  • FR
Paul Scherrer Institut (PSI)
Search
Paul Scherrer Institut (PSI)

Hauptnavigation

  • Research at PSIOpen mainmenu item
    • Research Initiatives
    • Ethics and Research integrity
    • Scientific Highlights
    • Scientific Events
    • Scientific Career
    • PSI-FELLOW
    • PSI Data Policy
  • Research Divisions and LabsOpen mainmenu item
    • Overview
    • Research with Neutrons and Muons
    • Photon Science
    • Energy and Environment
    • Nuclear Energy and Safety
    • Biology and Chemistry
    • Scientific Computing, Theory and Data
    • Large Research Facilities
  • Facilities and InstrumentsOpen mainmenu item
    • Overview
    • Large Research Facilities
    • Facilities
    • PSI Facility Newsletter
  • PSI User ServicesOpen mainmenu item
    • User Office
    • Methods at the PSI User Facilities
    • Proposals for beam time
    • Proposal Deadlines
    • Data Analysis Service (PSD)
    • EU support programmes
  • New ProjectsOpen mainmenu item
    • SLS 2.0
    • IMPACT
  • DE
  • EN
  • FR

Digital User Office (mobile)

  • Digital User Office

You are here:

  1. PSI Home
  2. Labs & User Services
  3. PSD
  4. LSC
  5. SMAM
  6. Scientific Highlights
  7. Deep learning-based monitoring of laser powder bed fusion processes

Secondary navigation

Structure and Mechanics of Advanced Materials

  • People
  • Research
    • Additive Manufacturing
    • Alignment in flow
    • Biological Tissue
    • In situ mechanical testing
    • Polymer Material
    • Tensor Tomography
  • Infrastructure
    • Selective Laser Melting
    • Deformation rigs
      • Micro Tensile Machine
      • Tension/Compression Module
      • Mini biaxial tensile machine
      • Meso biaxial tensile machine
      • Micro Shear Device
      • Micro Compression Device
  • Publications
  • Scientific Highlights
21 July 2022

Deep learning-based monitoring of laser powder bed fusion processes

A novel monitoring strategy for laser powder bed fusion processes based on a hybrid deep learning model is presented. It combines convolutional neural networks with long-short term memory. This model allows the classification of printing regimes with an accuracy of 98% across variable time-scales. 

LPBF Acoustic Emission
Miniaturized laser powder bed fusion device, equipped with an acoustic sensor in the build plate and an optical coherence collimator to collect optical emissions from the melt pool.

Harnessing the full potential of the metal-based Laser Powder Bed Fusion process (LPBF) relies heavily on how effectively the overall reliability and stability of the manufactured part can be ensured. To this aim, the recent advances in sensorization and processing of the associated signals using Machine Learning (ML) techniques have made in situ monitoring a viable alternative to post-mortem techniques such as X-ray tomography or ultrasounds for the assessment of parts. Indeed, the primary advantage of in situ monitoring over post-mortem analysis is that the process can be stopped in case of discrepancies, saving resources. Additionally, mitigations to repair the discrepancies can also be performed. However, the in situ monitoring strategies based on classifying processing regimes reported in the literature so far operate on signals of fixed length in time, constraining the generalization of the trained ML model by not allowing monitoring processes with heterogeneous laser scanning strategies.

Workflow machine learning
Workflow to build the dataset from operando experiments. Top: Schematics of the heterogeneous sensing system, comprising of photodiode detectors and AE sensor. Bottom: the schematic flow of the training strategy.

In this work, we bridge this gap by developing a hybrid Deep Learning (DL) model by combining Convolutional Neural Networks (CNNs) with Long-Short Term Memory (LSTM) that can operate over variable time-scales. The proposed hybrid DL model was trained on signals obtained from a heterogeneous time-synced sensing system consisting of four sensors, namely back reflection (BR), Visible, Infra-Red (IR), and structure-borne Acoustic Emission (AE). The signals captured different phenomena related to the LPBF process zone and were used to classify three regimes: Lack of Fusion (LoF), conduction mode and Keyhole. Specifically, these three regimes were induced by printing cubes out of austenitic Stainless steel (316L) on a mini-LPBF device with operando high-speed synchrotron X-ray imaging and signal acquisition with the developed heterogeneous sensing system. The operando X-ray imaging analysis ensured that the regimes correlated with the defined process parameters. During the validation procedure of the trained hybrid DL model, the model predicted three regimes with an accuracy of about 98% across various time scales, ranging from 0.5 ms to 4 ms. In addition to tracking the model performance, a sensitivity analysis of the trained hybrid model was conducted, which showed that the BR and AE sensors carried more relevant information to guide the decision-making process than the other two sensors used in this work.

Contact

Dr. Steven Van Petegem
Structure and Mechanics of Advanced Materials
Photon Science Division 
Paul Scherrer Institute
Forschungsstrasse 111, 5232 Villigen PSI, Switzerland
Telephone: +41 56 310 2537 
E-mail: steven.vanpetegem@psi.ch

Original publication

Deep Learning-based Monitoring of Laser Powder Bed Fusion Process on Variable Time-Scales Using Heterogeneous Sensing and operando X-ray Radiography Guidance 
Vigneashwara Pandiyan, Giulio Masinelli, Navarre Claire, Tri Le-Quang, Milad Hamidi-Nasab, Charlotte de Formanoir, Reza Esmaeilzadeh, Sneha Goel, Federica Marone, Roland Logé, Steven Van Petegem, Kilian Wasmer
Additive Manufacturing 58 (2022) 103007
DOI: 10.1016/j.addma.2022.103007

Sidebar

Contact

Structure and Mechanics of Advanced Materials Group

Paul Scherrer Institut
WBBA/114
5232 Villigen-PSI
Switzerland

Prof. Dr. Marianne Liebi
Group Leader
Telephone: +41 56 310 44 38
E-mail: marianne.liebi@psi.ch

Dr. Steven Van Petegem
Senior scientist
Telephone: +41 56 310 25 37
E-mail: steven.vanpetegem@psi.ch

LSC Homepage

Laboratory for Condensed Matter Physics


Photon Science Division

Homepage of PSI Division Photon Science (PSD)


Scientific Highlights PSD

Scientific Highlights of PSI Division Photon Science (PSD)

Swiss Light Source SLS

Synchrotron light large research facility.

SwissFEL

The new X-ray free electron laser facility.


User Office

The PSI User Office is a central PSI installation to serve the users from all the four user laboratories.


Current openings PSD

Job Opportunities at Research Division Photon Science

top

Footer

Paul Scherrer Institut

Forschungsstrasse 111
5232 Villigen PSI
Switzerland

Telephone: +41 56 310 21 11
Telefax: +41 56 310 21 99

How to find us
Contact

Visitor Centre psi forum
School Lab iLab (in German)
Center for Proton Therapy
PSI Education Centre
PSI Guest House
PSI Gastronomie (in German)
psi forum shop

Service & Support

  • Phone Book
  • User Office
  • Accelerator Status
  • PSI Publications
  • Suppliers
  • E-Billing
  • Computing
  • Safety (in German)

Career

  • Working at PSI
  • Job Opportunities
  • Training and further education
  • Career Center
  • Vocational Training (in German)
  • PSI Education Center

For the media

  • PSI in brief
  • Facts and Figures
  • Media corner
  • Media Releases
  • Social Media

Follow us: Twitter (in English) LinkedIn Youtube Facebook Instagram Issuu RSS

Footer legal

  • Imprint
  • Terms and Conditions
  • Editors' login