Skip to content

Deep learning-based topology optimization with a minimum compliance loss function

License

Notifications You must be signed in to change notification settings

elingaard/deep-topopt

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

deep-topopt

Deep learning-based topology optimization with a minimum compliance loss function.

Finite element meshing and analysis is handled in the FEA module, while the loss function and analytical gradients are handled in the spalg module.

To run the model type python main.py 'SAVE_PATH' 'MODEL_NAME' 'DATA_PATH'. The following additional arguments can also be parsed:

parser.add_argument('--pretrained',default="") # save path for pretrained weights
parser.add_argument('--batch_size', default=16, type=int) 
parser.add_argument('--epochs', default=100, type=int)
parser.add_argument('--lr', default=2e-4, type=float) # learning rate
parser.add_argument('--clip', default=1, type=float) # gradient clipping value
parser.add_argument('--dens_penal', default=2, type=float) # SIMP penalization
parser.add_argument('--vol_penal', default=1e1, type=float) # volume penalization

alt text

Send a mail to maoel@mek.dtu.dk if you want to obtain the datasets needed to train the model.

About

Deep learning-based topology optimization with a minimum compliance loss function

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages