This is the Github repository for the paper "Efficient Water Desalination with Graphene NanoporesObtained using Artificial Intelligence". In this work, we propose a graphene nanopore optimization framework via the combination of DRL and CNN for efficient water desalination. The DRL agent controls the growth of nanopore, while the CNN is employed to predict the water flux and ion rejection of the nanoporous graphene membrane at a certain external pressure. Experiments show that our framework can design nanopore structures that are promising in energy-efficient water desalination.
- Windows, Linux or macOS
- Python 3
- CPU or NVIDIA GPU + CUDA CuDNN
$ git clone https://github.com/yuyangw/Graphene-RL-DQN.git
$ cd Graphene-RL-DQN
$ conda env create --name graphene-rl --file env.yml
$ conda activate graphene-rl
$ python cnn.py
$ python main.py
Here we show nanopore evolution controlled by the DRL agent
The data used to train the CNN water desalination performance predictor is stored in the data folder. Data include the PDB files of each graphene nanopore and it's augmented variants, and a csv file that record their corresponding values of water flux and ion rejection rate. Images of nanopores can be generated by
pdb2img.py