A Fast Learning-Based Unit Commitment Strategy with AC Optimal Power Flow for Large Grids with Direct Inclusion of Weather
Farnaz Safdarian, Joshua Peeples, Dillon Richards, Jessica Wert, Thomas Overbye
Note: If this code is used, cite it: Farnaz Safdarian, Joshua Peeples, Dillon Richards, Jessica Wert, Thomas Overbye. (2023, November 9) Peeples-Lab/UCML: Initial Release (Version v1.0).
In this repository, we provide the paper and code for "A Fast Learning-Based Unit Commitment Strategy with AC Optimal Power Flow for Large Grids with Direct Inclusion of Weather."
The requirements.txt
file includes all the necessary packages, and the packages will be installed using:
pip install -r requirements.txt
We strongly recommend creating a virtual environment
to install necessary packages.
To get started, please follow the instructions in the Datasets and Pretrained_Models folders to download the dataset(s) and model(s) respectively.
Next, run demo.py
in Python IDE (e.g., Spyder) or command line to train, validate, and test models.
The UC ML code uses the following functions.
-
Load data
features, labels = load_dataset(**Parameters)
-
Prepare dataset(s) for model (train/validation/test split)
X_train, X_test, y_train, y_test = train_test_split(**Parameters)
-
Cross validate model(s) for hyperparameter tuning
classifiers = cross_validation(**Parameters)
-
Retrain best model(s) with full training data, test model, and get quantitative and qualitative results
generate_PR_Curves(**Parameters)
The parameters can be set on the command line and the parameters for the experiments are stored in the Demo_Parameters.py
.
https://github.com/Advanced-Vision-and-Learning-Lab/UC_ML
└── root dir
├── demo.py //Run this. Main demo file.
├── Demo_Parameters.py // Parameter file for the demo.
└── Utils //utility functions
├── crossval.py // Contains functions perform hyperparameter tuning for each classifier.
├── PR_curves.py // Generate precision-recall curves for each classifier.
This source code is licensed under the license found in the LICENSE
file in the root directory of this source tree.
This product is Copyright (c) 2023 F. Safdarian, J. Peeples, D. Richards, J. Wert, T. Overbye. All rights reserved.
If you use the Unit Commitment Machine Learning code, please cite the following reference using the following entry.
Plain Text:
F. Safdarian, J. Peeples, D. Richards, J. Wert, and T. Overbye, "A Fast Learning-Based Unit Commitment Strategy with AC Optimal Power Flow for Large Grids with Direct Inclusion of Weather," in Review.
BibTex:
@inproceedings{Safdarian2023fast,
title={A Fast Learning-Based Unit Commitment Strategy with AC Optimal Power Flow for Large Grids with Direct Inclusion of Weather},
author={Safdarian, Farnaz and Peeples, Joshua, and Richards, Dillon, and Wart, Jessica, and Overbye, Thomas},
booktitle={TBD},
pages={TBD},
year={2023},
organization={TBD}
}