[IEEE Transactions on Power Systems] Transmission Interface Power Flow Adjustment: A Deep Reinforcement Learning Approach based on Multi-task Attribution Map

Official codebase for paper Transmission Interface Power Flow Adjustment: A Deep Reinforcement Learning Approach based on Multi-task Attribution Map. This codebase is based on the open-source Tianshou and PandaPower framework and please refer to those repo for more documentation.

A novel approach named as FSA is recently proposed to solve the same task as an enhancement to MAM.

Overview

TLDR: This work is the first dedicated attempt towards learning multiple transmission interface power flow adjustment tasks jointly, a highly practical problem yet largely overlooked by existing literature in the field of the power system. We design a novel deep reinforcement learning (DRL) method based on multi-task attribution map (MAM) to handle multiple adjustment tasks jointly, where MAM enables the DRL agent to selectively integrate the node features into a compact task-adaptive representation for the final adjustment policy. Simulations are conducted on the IEEE 118-bus system, a realistic 300-bus system in China, and a very large European 9241-bus system, demonstrating that the proposed method brings remarkable improvements to the existing methods. Moreover, we verify the interpretability of the learnable MAM in different operation scenarios.

Prerequisites

Install dependencies

python==3.8.13
dgl==1.1
torch==1.13
pandapower==2.11
gym==0.23
tianshou==0.4.11
numpy==1.22.4
numba==0.55.2
pandas==1.4.2

Usage

Please follow the instructions below to replicate the results in the paper. Note that the model of the realistic 300-bus system in China is not available due to confidentiality policies of SGCC.

Unzip the data for training and testing under the same working directory:

tar -Jxvf data.tar.xz

Train the DRL agent with MAM:

# IEEE 118-bus System under the multi-task setting (10 single-interface tasks)
python train.py --case='case118' --task='S10' --method='MAM' --model='Attention'

# IEEE 9241-bus System under the multi-task setting (10 single-interface tasks)
python train.py --case='case9241' --task='S10' --method='MAM' --model='Attention'

# IEEE 118-bus System under the multi-task setting (different 5-interface tasks)
python train.py --case='case118' --task='M5' --method='MAM' --model='Attention'

# IEEE 9241-bus System under the multi-task setting (different 3-interface tasks)
python train.py --case='case9241' --task='M3' --method='MAM' --model='Attention'

Citation

If you find this work useful for your research, please cite our paper:

@article{liu2023MAM,
  author={Liu, Shunyu and Luo, Wei and Zhou, Yanzhen and Chen, Kaixuan and Zhang, Quan and Xu, Huating and Guo, Qinglai and Song, Mingli},
  journal={IEEE Transactions on Power Systems}, 
  title={Transmission Interface Power Flow Adjustment: A Deep Reinforcement Learning Approach Based on Multi-Task Attribution Map}, 
  year={2023},
  volume={},
  number={},
  pages={1-12},
  doi={10.1109/TPWRS.2023.3298007}
}

Contact

Please feel free to contact me via email (liushunyu@zju.edu.cn, davidluo@zju.edu.cn) if you are interested in my research :)

Zzhangquan / MAM