A deep learning approach to estimate stress distribution: a fast and accurate surrogate of finite-element analysis

The files contain code and data associated with the paper titled "A Deep Learning Approach to Estimate Stress Distribution: A Fast and Accurate Surrogate of Finite Element Analysis". The paper is authored by Liang Liang, Minliang Liu, Caitlin Martin, and Wei Sun, and published at Journal of The Royal Society Interface, 2018.

The following repository contains further investigation of the method used in the mentioned above paper.

Copyright (c) 2018 by Biologically Inspired Computer Vision Group at Imperial College London. All rights reserved.

1. Data:

• ShapeData.mat
• StressData.mat

2. Code of DL-model:

• DLStress.py
• im2patch.m
• UnsupervisedLearning.m

3. Code for visualization:

• ReadMeshFromVTKFile.m
• ReadPolygonMeshFromVTKFile.m
• WritePolygonMeshAsVTKFile.m
• Visualization.m

4. Template meshes for visualization:

• TemplateMesh3D.vtk
• TemplateMesh2D.vtk

• OS: Windows (64bit) 7 or 10
• Hardware: Intel quad-core CPU, 32G RAM

• Anaconda 3.6: https://www.anaconda.com/download/;
• Keras 2.0.4: https://github.com/fchollet/keras. Keras can be install from Anaconda Cloud: https://anaconda.org/anaconda/keras
• Tensorflow 1.1.0: https://www.tensorflow.org/. Tensorflow CPU version can be installed from Anaconda Cloud: https://anaconda.org/conda-forge/tensorflow
• Matlab 2017a: https://www.mathworks.com/products/matlab.html
• MatConvNet: http:https://www.vlfeat.org/matconvnet/; version 1.0-beta24 (backward compatibility not guaranteed)
• Paraview: https://www.paraview.org/download/; https://www.paraview.org/paraview-guide/
• Spyder: https://spyder-ide.github.io
• Python 3.5

1. Install Matlab
2. Install MatConvNet - In Matlab software go to directory ...\matconvnet-1.0-beta25\matlab\ and run the file:

v1_compilenn.m

3. Install Anaconda
4. Make downgrade to python version 3.5 using following formula

conda install python=3.5

5. Install Tensorflow in Anaconda

conda create -n tensorflow pip python=3.5
activate tensorflow
pip install --ignore-installed --upgrade tensorflow

6. Install keras and then change the keras_backend to Tensorflow (in the path: .../anaconda3/envs/tensorflow/etc/conda/activate.d/keras_activate.sh)

conda install -c conda-forge keras

7. Install Spyder in Anaconda: https://anaconda.org/conda-forge/spyder
8. Setup Matlab engine for python

c:\matlab\R2017a\extern\engines\python
python setup.py install

1. Activate the anaconda environment in a cmd window, and type spyder. Then you should see something like this. Spyder is a Python IDE. The current directory of Spyder is shown on top right. Open DLStress.py in Spyder, and run the code. You need to change the current directory of Spyder so that it contains DLStress.py. Change the path of MatConvnet in UnsupervisedLearning.m
2. Create in the main folder empty directory: 'result' and 'result2'
3. Once you save the result to mat files, open Visualization.m, and then convert the result to vtk files.
4. Open the vtk files in Paraview. You will see the ground-truth and predicted stress fields on 2D/3D meshes.