Abstract—Efficient management of energy resources relies heavily on residential electricity usage. However, challenges such as defective smart meters, erratic consumption patterns, and energy theft pose serious threats to the stability of the smart grid. Addressing these irregularities requires robust anomaly detection methods. This study presents a novel approach to anomaly detection in smart grid systems by integrating Generative Adversarial Networks (GANs) and Convolutional Neural Networks (CNNs). The GAN generates synthetic electricity consumption data closely resembling actual usage patterns, leveraging its understanding of typical consumption behaviors learned from extensive historical data. Meanwhile, a CNN architecture extracts crucial features from consumption data, capturing spatial and temporal correlations. Through a rigorous training process, the GAN-CNN model learns to identify regular consumption patterns and detect anomalies effectively. Extensive testing using real-time consumption data demonstrates the model’s remarkable accuracy, achieving an impressive accuracy rate of approximately 0.9976. This methodology proves robust, exhibiting high detection accuracy, low false positive rates, and resilience to noise and data volatility. Implementation of this GAN-CNN-based anomaly detection approach in smart grid systems holds significant implications for enhancing grid security, resource distribution efficiency, and customer service. It empowers utilities to swiftly address equipment issues and mitigate energy theft, thereby improving overall grid stability and operational effectiveness. Consequently, this research contributes substantially to the advancement of smart grid analytics and sets the stage for future innovations in the field.
 
Keywords—Generative Adversarial Network (GAN), Convolutional Neural Network (CNN), anomaly detection


Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).