yubo-huang/maobi

Code for the paper 'Wind Farm Control via Offline Reinforcement Learning with Adversarial Training'

Multi-agent offline behavior clone

This codebase is the implementation of paper named Wind Farm Control via Offline Reinforcement Learning with Adversarial Training.

1. ./data directory contains the training samples of 8 wind turbines
2. ./learn directory saved the learned control policy

How to train MAOBI

First, cloning this repo to your local computer using

git clone https://github.com/Yubo-Huang/maobi.git

cd maobi

Then, running the .sh file:

./run.sh