AUTHOR=Duan Jing 

TITLE=Deep learning anomaly detection in AI-powered intelligent power distribution systems

JOURNAL=Frontiers in Energy Research

VOLUME=Volume 12 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/energy-research/articles/10.3389/fenrg.2024.1364456

DOI=10.3389/fenrg.2024.1364456

ISSN=2296-598X

ABSTRACT=Intelligent power distribution systems play a crucial role in modern power industries, managing power distribution effectively. However, current systems face challenges in anomaly detection due to limitations in data complexity and model generalization. To address this, we propose a Transformer-GAN model that combines deep learning and generative adversarial network (GAN) technology to enhance anomaly detection. Our model utilizes the Transformer's self-attention mechanism to process high-dimensional, complex time series data, improving the identification of abnormal patterns in power systems. The GAN component enhances the model's robustness and adaptability to dynamic changes and unknown abnormalities. Experimental results demonstrate the superior performance of our model over traditional methods on multiple datasets, offering an advanced solution for power system anomaly detection. This research has practical significance in enhancing the stability and security of smart power distribution systems. It also has potential applications in industrial automation and the Internet of Things, contributing to overall system performance and maintainability. Our study supports the application of artificial intelligence in the power field, paving the way for the reliability and intelligent development of future power systems.