jiangyuan-li / deep-spatio-temporal

Code for Deep Spatio Temporal Wind Power Forecasting

Deep-Spatio-Temporal

Code for Deep Spatio-Temporal Wind Power Forecasting

How to use

The model is validated on two datasets.

Wind power forecasting

This dataset is from https://aml.engr.tamu.edu/book-dswe/dswe-datasets/. The data used here is Wind Spatio-Temporal Dataset2. Download data, put it into the './data' folder and rename it to 'wind_power.csv'. Then, run following

python train.py --name wind_power --epoch 300 --batch_size 20000 --lr 0.001 --k 5 --n_turbines 200

Wind speed forecasting

The model performance on wind speed forecasting is validated on NREL WIND dataset (https://www.nrel.gov/wind/data-tools.html). We select one wind farm with 100 turbines from Wyoming. To get data, first run

python getNRELdata.py

Then run

python train.py --name wind_speed --epoch 300 --batch_size 20000 --lr 0.001 --k 9 --n_turbines 100

References

• Jiangyuan Li, Mohammadreza Armandpour. (2022) "Deep Spatio-Temporal Wind Power Forecasting". IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

deep-spatio-temporal's Issues

您好，对比模型的代码可以添加上吗？比如PSTN,STAN

about mae and rmse

hello， I would like to ask that when calculating the rating indicators, y_ true and y_ preddid pred add 1 and divide by 2,thank you.

Doubt about inputs

Hello, we wanted to test your algorithm with our data. Is a windfarm in France.

We have some doubts:

1. Our wind speed historical data are not exactly in the same location as tubrines locations. But we can update de code in a way that for each turbine the algorithm take the closer wind speed locations?
2. Our wind speed historical data are in vector form (u,v). Should we do the module of the vector to get the same speed as your data, or maybe we could try te algorithm with our data directly, so for each location we have two values, the componente u and v.
   Thanks for your attention.

Regards

a little pardon plz

mj

How can I see the results

I have finished the train,run the train.py and getNRELdata.py,but nothng happens, I don't know how to see the results. Is this result a model?? In a word,I don't know what is the output .I have seen the article,but I can't understand.Because I am a new in this field and I can't find others help me.Thank you so much for your help

Data about 100 wind turbines in Wyoming

When run getNRELdata.py on my machine， it shows "ValueError: no endpoint set". I think the URL:"/nrel/wtk/conus/wtk_conus_2012.h5" is not available right now. I want to know how can I get this dataset， thanks a lot！