adaptive-spreading

Code for the paper "Learning Controllers for Adaptive Spreading of Carbon Fiber Tows". Download the data (preprocessed) and a pretrained process model here: https://figshare.com/s/1a3e9b1ac16362b46cf9

Run experiments

• Download the data and adjust the data path if necessary (and other in-/output paths) in the utils/paths.py file.
  The data is used for both, the Process Model and the Process Control Model.

Process Model

• To train a feedforward neural network or a random forest, run python -m tape_prediction.start_ff or python -m tape_prediction.start_rf

• Results can be found in the generated log file (per default in the logs directory)

• When training a neural network, in addition, a tensorboard file is generated (in the runs directory).

Process Control Model

• To start the neuroevolution, run python -m process_control_model.start_neuroevolution

  • Parameters can be set in the source code or passed as json-file (see process_control_model/params/example_ne_param_file.json for details)
  • Please adjust the type of the process_model used as backend (i.e. "nn" or "rf") and specify the path to the pre-trained model.
  • Results are logged to a file in the directory process_control_model/logs

• The baseline can be run as follows python -m process_control_model.fixed_setup. Log files can be found in the same directory as the neuroevolution logs.

General Remarks

• See requirements.txt for libraries required to run the code.
• Additional folders are created in the project directory (according to the paths specified in utils/paths.py) when necessary.