Learning of E-Jet Printing Dynamics with Physics-Informed Gaussian Processes

This Github repository contains all the data and code needed to replicate the results reported in this paper:

Oikonomou et al. (2023), "Physics-Informed Bayesian learning of electrohydrodynamic polymer jet printing dynamics", Communications Engineering, Nature

GPJet Code

Learn how to set up the environment, run and use the code: HERE

To run the GPJet framework download or clone the github repo and install packages as referred at gpjet.yml

Otherwise, open anaconda prompt and run:

conda env create --file gpjet.yml

Part I: Computer Vision Module

Description:
A real time computer vision framework using OpenCV for real-time feature extraction relevant to MEW process dynamics.

• code: LINK
  • sequential
  • concurrent
  • parallel
• data: LINK
  • initially processed videos for JM framework (kommena ta aspra panw katw)
• results: LINK
  • processed videos after JM framework
  • these videos were used for jet feature extraction

Part II: Machine Learning Module

Description:
Surrogate modeling using jet metrology data.

• code: LINK
  • GP Regression (GPR) -> Surrogate Models
  • Multi-Fidelity (MFD) -> Surrogate Models & Physics Models
  • Active Learning 1 (AL1) -> Surrogate Models
  • Active Learning 2 (AL2) -> Multi-Fidelity Models
• dataset:
  • GPR & AL1: LINK
  • MFD & AL2:
    • High fidelity (extracted jet features) dataset from Computer Vision Module:
      • LINK
    • Low fidelity dataset from Jet Diameter Model:
      • LINK
• results: LINK

Part III: Physics Module

Description:

• Jet Diameter Model: LINK
  • Results: LINK
• Mechanical Fluid Sewing Machine Patterns: LINK
  • Results: LINK