This repo accompanies a paper submitted to Additive Manufacturing in 2024 and contains a reproduction of the methodology demonstrated in that paper.
The framework of simulation-in-the-loop real-time FEA product structural quality validation system.
The defect segmentation process with U-net.
The process of real-time automatic FEA structural quality validation on the fourth layer.
- Python 3.8.5
- Abaqus 2020
- numpy >= 1.20
- tensorflow
- scikit-learn
- scikit-image
- onnxruntime
- onnx
- tf2onnx
- skl2onnx
- opencv-python
- overrides
- dataclasses-json
- matplotlib
- SciencePlots
- segmentation_models
- watchdog
The provided code has the following folder structure, where the files are structured based on the stage of the pipeline.
- Code
- FEA
- final_3d_printer_fea
- Data
- Defect segmentation results
- DIC test results
- Specimen
The Code folder contains the two parts: 1) the generated Abaqus Python Script from the simulation and 2) the code for running the FEA simulation.
The Data folder contains the two parts: 1) Defect segmentation results and 2) DIC test results.
The Specimen folder is the Gcode for the specimen used in the paper
- cd .....\final_3d_printer_fea
- conda activate fea_printing(Virtual Environment)
- python main.py
- Replacing the trained image segmentation result (here: New_specimen_with_prosity_and_long_gap.hdf5 in test folder) with your own.
- Pay attention to the image size, in the paper we use 1368 X 1020, which is resized to 512 X 512 before input into U-net.
- Replacing the pregenerated Abaqus Python Script (here: Abaqus_simulation.py ) with your own.
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Cite as:
@Misc{Fu_2024_SimulationintheLoop,
author = {Yanzhou Fu and Austin R.J. Downey},
howpublished = {GitHub},
title = {Paper-Simulation-in-the-loop-Additive-Manufacturing-for-Real-time-Structural-Validation},
year = {2024},
groups = {{ARTS-L}ab},
url = {https://github.com/ARTS-Laboratory/Paper-Simulation-in-the-loop-Additive-Manufacturing-for-Real-time-Structural-Validation},
}