Mohammad-Abazari / large-scale-truss-optimization

Large scale truss optimization using NSGA-II

Optimizing weight and compliance of a large 3D truss

Multi-objective optimization of the weight and compliance of a large 3D truss subject to stress and displacement constraints [1] using NSGA-II [3]. Pymoo [2] is used for implementing NSGA-II. Finite element analysis (FEA) is performed in order to calculate the truss compliance. The FEA code is available under both Python and MATLAB.

Running unit tests to ensure everything is in order:

1. Clone the repo:

```https://github.com/abhiroopghosh71/large-scale-truss-optimization.git```

2. Change into the working directory cd large-scale-truss-optimization

3. To install the dependencies use pip install requirements.txt. If you are using Anaconda, it is recommended to create a new virtual environment using conda create --name <envname> python=3.8 --file requirements.txt

4. OPTIONAL: Install the MATLAB Engine API for Python if the MATLAB FEA code is preferred. Instructions can he found on the MathWorks website

5. Change into the tests directory cd tests

6. To test the Python codes run: pytest test_truss_python_fea.py and pytest test_truss_pymoo_python_fea_parallel.py

7. If MATLAB Engine API was installed previously, run the tests: pytest test_truss_matlab_fea.py and pytest test_truss_pymoo_matlab_fea.py

Running the optimization

1. Change into the large-scale-truss-optimization directory.

2. To run the optimization use: python optimize.py [OPTIONS]. For example, to run the optimization with a 40 population size and 100 generations, run: python optimize.py --popsize 40 --ngen 100

Important command line optimization parameters

1. --seed <value>: Sets the seed for the random number generator.

2. --ngen <value>: Number of generations of NSGA-II.

3. --popsize <value>: Population size of NSGA-II.

4. --nshapevar <value>: Number of shape variables [1].

5. --symmetric: Enforcing symmetry in the truss [1].

Please report issues to me, Abhiroop Ghosh, at [e-mail].

References:

1. A. Ghosh, K. Deb, R. Averill, E. Goodman, "Combining User Knowledge and Online Innovization for Faster Solution to Multi-objective Design Optimization Problems", https://doi.org/10.1007/978-3-030-72062-9_9

2. J. Blank and K. Deb, pymoo: Multi-Objective Optimization in Python, in IEEE Access, vol. 8, pp. 89497-89509, 2020, https://doi.org/10.1109/ACCESS.2020.2990567

3. K. Deb, A. Pratap, S. Agarwal and T. Meyarivan, "A fast and elitist multiobjective genetic algorithm: NSGA-II," in IEEE Transactions on Evolutionary Computation, vol. 6, no. 2, pp. 182-197, April 2002. https://doi.org/10.1109/4235.996017