Implementation of TOuNN paper TOuNN: Topology Optimization using Neural Networks in MATLAB
The original repository for the python implementation of the paper https://github.com/UW-ERSL/TOuNN
by the authors
Aaditya Chandrasekhar, Krishnan Suresh, Engineering Representations and Simulation Lab, University of Wisconsin-Madison
There are some changes in this implementation namely:
- FEA is based on 3D Hexahedral elements
- Has 3 types of gradient thresholding to choose from
- Heavily inspired from top3D125 and top3d
- The number of trainable parameters of the Neural Network is dependent on the number of elements but with 5 hidden layers.
- Multiple simultaneous loads
Cantilever with point load.
TOuNN(70, 30, 2, 0.5);Will display only the final design.
TOuNN(70, 30, 2, 0.5, true);Will produce the following gif but will take lot of time. Not Recommeneded.
For any changes in loads and/or boundary conditions, make appropriate changes in the problem_def.m file before calling the TOuNN function
