Data efficiency assessment of generative adversarial networks in energy applications

Umme Mahbuba Nabila; Linyu Lin; Xingang Zhao; William L. Gurecky; Pradeep Ramuhalli; Majdi I. Radaideh

doi:10.1016/j.egyai.2025.100501

Data efficiency assessment of generative adversarial networks in energy applications

Umme Mahbuba Nabila, Linyu Lin, Xingang Zhao, William L. Gurecky, Pradeep Ramuhalli, Majdi I. Radaideh

Modern Nuclear I&C

Research output: Contribution to journal › Article › peer-review

6 Scopus citations

Abstract

This study investigates the data requirements of generative artificial intelligence (AI), particularly generative adversarial networks (GANs), for reliable data augmentation in energy applications. Generative AI, though seen as a solution to data limitations, requires substantial data to learn meaningful distributions—a challenge often overlooked. This study addresses the challenge through synthetic data generation for critical heat flux (CHF) and power grid demand, focusing on renewable and nuclear energy. Two variants of GAN employed are conditional GAN (cGAN) and Wasserstein GAN (wGAN). Our findings include the strong dependency of GAN on data size, with performance declining on smaller datasets and varying performance when generalizing to unseen experiments. Mass flux and heated length significantly influence CHF predictions. wGAN is more robust to feature exclusion, making it suitable for constrained synthetic data generation. In energy demand forecasting, wGAN performed well for solar, wind, and load predictions. Longer lookback hours and larger datasets improved predictions, especially for load power. Seasonal variations posed challenges, with wGAN achieving a relatively high error of Root Mean Squared Error (RMSE) of 0.32 for load power prediction, compared to RMSE of 0.07 under same-season conditions. Feature exclusions impacted cGAN the most, while wGAN showed greater robustness. This study concludes that, while generative AI is effective for data augmentation, it requires substantial data and careful training to generate realistic synthetic data and generalize to new experiments in engineering applications.

Original language	English
Article number	100501
Journal	Energy and AI
Volume	20
DOIs	https://doi.org/10.1016/j.egyai.2025.100501
State	Published - May 2025

Funding

The first author would like to thank Farah Alsafadi from North Carolina State University for the valuable discussions about generative models at the beginning of this project. This work was supported through Idaho National Laboratory, United States's Laboratory Directed Research and Development (LDRD) Program Award Number (24A1081-116FP) under Department of Energy Idaho Operations Office contract no. DE-AC07-05ID14517. The authors also acknowledge the use of Idaho National Laboratory, United States's high performance computing (HPC), for providing computational resources, which significantly contributed to the modeling and analysis presented in this work. Additionally, it was partially funded by the AI Initiative as part of the Laboratory Directed Research and Development Program of Oak Ridge National Laboratory (ORNL), United States, which is managed by UT-Battelle LLC for the US Department of Energy under contract DE-AC05-00OR22725. This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan. The first author would like to thank Farah Alsafadi from North Carolina State University for the valuable discussions about generative models at the beginning of this project. This work was supported through Idaho National Laboratory\u2019s Laboratory Directed Research and Development (LDRD) Program Award Number (24A1081-116FP) under Department of Energy Idaho Operations Office contract no. DE-AC07-05ID14517. The authors also acknowledge the use of Idaho National Laboratory\u2019s high performance computing (HPC), for providing computational resources, which significantly contributed to the modeling and analysis presented in this work. Additionally, it was partially funded by the AI Initiative as part of the Laboratory Directed Research and Development Program of ORNL , which is managed by UT-Battelle LLC for the US Department of Energy under contract DE-AC05-00OR22725 .

Keywords

Critical heat flux
Data augmentation
Generative adversarial networks
Generative AI
Power grid energy forecasting

Access to Document

10.1016/j.egyai.2025.100501

Cite this

@article{33d69a61db694614969c647854a7d68c,

title = "Data efficiency assessment of generative adversarial networks in energy applications",

abstract = "This study investigates the data requirements of generative artificial intelligence (AI), particularly generative adversarial networks (GANs), for reliable data augmentation in energy applications. Generative AI, though seen as a solution to data limitations, requires substantial data to learn meaningful distributions—a challenge often overlooked. This study addresses the challenge through synthetic data generation for critical heat flux (CHF) and power grid demand, focusing on renewable and nuclear energy. Two variants of GAN employed are conditional GAN (cGAN) and Wasserstein GAN (wGAN). Our findings include the strong dependency of GAN on data size, with performance declining on smaller datasets and varying performance when generalizing to unseen experiments. Mass flux and heated length significantly influence CHF predictions. wGAN is more robust to feature exclusion, making it suitable for constrained synthetic data generation. In energy demand forecasting, wGAN performed well for solar, wind, and load predictions. Longer lookback hours and larger datasets improved predictions, especially for load power. Seasonal variations posed challenges, with wGAN achieving a relatively high error of Root Mean Squared Error (RMSE) of 0.32 for load power prediction, compared to RMSE of 0.07 under same-season conditions. Feature exclusions impacted cGAN the most, while wGAN showed greater robustness. This study concludes that, while generative AI is effective for data augmentation, it requires substantial data and careful training to generate realistic synthetic data and generalize to new experiments in engineering applications.",

keywords = "Critical heat flux, Data augmentation, Generative adversarial networks, Generative AI, Power grid energy forecasting",

author = "Nabila, \{Umme Mahbuba\} and Linyu Lin and Xingang Zhao and Gurecky, \{William L.\} and Pradeep Ramuhalli and Radaideh, \{Majdi I.\}",

note = "Publisher Copyright: {\textcopyright} 2025 The Authors",

year = "2025",

month = may,

doi = "10.1016/j.egyai.2025.100501",

language = "English",

volume = "20",

journal = "Energy and AI",

issn = "2666-5468",

publisher = "Elsevier",

}

TY - JOUR

T1 - Data efficiency assessment of generative adversarial networks in energy applications

AU - Nabila, Umme Mahbuba

AU - Lin, Linyu

AU - Zhao, Xingang

AU - Gurecky, William L.

AU - Ramuhalli, Pradeep

AU - Radaideh, Majdi I.

PY - 2025/5

Y1 - 2025/5

N2 - This study investigates the data requirements of generative artificial intelligence (AI), particularly generative adversarial networks (GANs), for reliable data augmentation in energy applications. Generative AI, though seen as a solution to data limitations, requires substantial data to learn meaningful distributions—a challenge often overlooked. This study addresses the challenge through synthetic data generation for critical heat flux (CHF) and power grid demand, focusing on renewable and nuclear energy. Two variants of GAN employed are conditional GAN (cGAN) and Wasserstein GAN (wGAN). Our findings include the strong dependency of GAN on data size, with performance declining on smaller datasets and varying performance when generalizing to unseen experiments. Mass flux and heated length significantly influence CHF predictions. wGAN is more robust to feature exclusion, making it suitable for constrained synthetic data generation. In energy demand forecasting, wGAN performed well for solar, wind, and load predictions. Longer lookback hours and larger datasets improved predictions, especially for load power. Seasonal variations posed challenges, with wGAN achieving a relatively high error of Root Mean Squared Error (RMSE) of 0.32 for load power prediction, compared to RMSE of 0.07 under same-season conditions. Feature exclusions impacted cGAN the most, while wGAN showed greater robustness. This study concludes that, while generative AI is effective for data augmentation, it requires substantial data and careful training to generate realistic synthetic data and generalize to new experiments in engineering applications.

AB - This study investigates the data requirements of generative artificial intelligence (AI), particularly generative adversarial networks (GANs), for reliable data augmentation in energy applications. Generative AI, though seen as a solution to data limitations, requires substantial data to learn meaningful distributions—a challenge often overlooked. This study addresses the challenge through synthetic data generation for critical heat flux (CHF) and power grid demand, focusing on renewable and nuclear energy. Two variants of GAN employed are conditional GAN (cGAN) and Wasserstein GAN (wGAN). Our findings include the strong dependency of GAN on data size, with performance declining on smaller datasets and varying performance when generalizing to unseen experiments. Mass flux and heated length significantly influence CHF predictions. wGAN is more robust to feature exclusion, making it suitable for constrained synthetic data generation. In energy demand forecasting, wGAN performed well for solar, wind, and load predictions. Longer lookback hours and larger datasets improved predictions, especially for load power. Seasonal variations posed challenges, with wGAN achieving a relatively high error of Root Mean Squared Error (RMSE) of 0.32 for load power prediction, compared to RMSE of 0.07 under same-season conditions. Feature exclusions impacted cGAN the most, while wGAN showed greater robustness. This study concludes that, while generative AI is effective for data augmentation, it requires substantial data and careful training to generate realistic synthetic data and generalize to new experiments in engineering applications.

KW - Critical heat flux

KW - Data augmentation

KW - Generative adversarial networks

KW - Generative AI

KW - Power grid energy forecasting

UR - https://www.scopus.com/pages/publications/105001477295

U2 - 10.1016/j.egyai.2025.100501

DO - 10.1016/j.egyai.2025.100501

M3 - Article

AN - SCOPUS:105001477295

SN - 2666-5468

VL - 20

JO - Energy and AI

JF - Energy and AI

M1 - 100501

ER -

Data efficiency assessment of generative adversarial networks in energy applications

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this