warwick-icse/SFNet
Multi-fidelity modeling of wind farm wakes based on a novel super-fidelity network
Multi-fidelity modeling of wind farm wakes based on a novel super-fidelity network
Energy Conversion and Management, 2022, Rui Li, Jincheng Zhang and Xiaowei Zhao.
Introduction
This project (SFNet) is the pretrained model and test code for Multi-fidelity modeling of wind farm wakes based on a novel super-fidelity network.
Preparation
python==3.6.13
torch==1.10.2
torchvision==0.11.3
floris==2.4
pandas==1.1.5
numpy==1.19.5
Visualization
After installing required libraries mentioned above, then you can run the test.py based on provided low-fidelity flow fields generated by FLORIS (8 m/s, 9 m/s and 10 m/s). We provide three pretrained models which are trained based on 45, 90 and 135 samples (you can choose different models by changing the value of pre_trained_sample in test.py). To test different wind speeds, you need to change of the value of wind_speed in test.py.
Or you can generate your own data using gene_floris_farm.py.
Results
The flow fields generated by FLORIS (left) and enhanced by SFNet (right):
Citation
If you find this project useful in your research, please consider citing our paper：
@Article{
li2022multi,
title={Multi-fidelity modeling of wind farm wakes based on a novel super-fidelity network},
author={Li, Rui and Zhang, Jincheng and Zhao, Xiaowei},
journal={Energy Conversion and Management},
volume={270},
pages={116185},
year={2022},
publisher={Elsevier}
}