RL for combustion engine

Environments and scripts for doing reinforcement learning for injections into different types of engines.

Running the tests

$ conda env create -f environment.yaml
$ nosetests

Getting started

To train an agent, you can run the PPO example:

$ cd scripts/ppo-example
$ python ../main.py --fname input.toml

Citing

@article{Henrydefrahan2021,
  title={Deep reinforcement learning for dynamic control of fuel injection timing in multi-pulse compression ignition engines},
  author={Henry de Frahan, Marc T and Wimer, Nicholas T and Yellapantula, Shashank and Grout, Ray W},
  journal={International Journal of Engine Research},
  volume={-},
  pages={-},
  year={2021},
  publisher={Sage}
}