Project carried out as a visiting doctoral student with the Sustainable Systems Design Lab at the University of Victoria, Canada, to apply machine learning for the optimisation of a offshore direct air capture (DAC) system. See the Wiki for modelling, control algorithms and results.

First, create a conda environment and install the required packages:

conda create -n dac_control
conda activate dac_control
pip install -r requirements.txt

Weights and Biases is used to track the experiments. After signing up for a free account, run the following:

wandb login
wandb online

Hydra is used to manage the experiment configuration using config files. To run an experiment with the default configuration (as stated in main.yaml), run:

python system_simulation.py

The experiment configurations can be changed by specifying the file name or overriden directly, for example:

python system_simulation.py dac.num_units=6 controller=cycling kinetics=sizing_nums_dacs

To specify the cross entropy method (CEM) model predictive controller (MPC), a dynamics model must be specified in the command line, for example:

python system_simulation.py controller=cem +dynamics_model=constant_wind

A seperate script can be used to train and save a reinforcement learning agent using Stable-Baselines3.

python train_rl_agent.py controller=rl_ac

After training the agent, the learnt policy can be evaluated by running:

python system_simulation.py controller=rl_ac +controller.name='W&B_NAME'