Quantum Autoencoder-Based Anomaly Detection for Cyberattacks in Smart Power Systems

The growing complexity and interconnectivity of modern power systems have increased their vulnerability to sophisticated cyberattacks, making real-time anomaly detection a critical challenge. Classical machine learning approaches, while effective, face limitations in scalability and representation power, especially when dealing with high-dimensional and temporally correlated data. Quantum autoencoders (QAEs) offer a promising alternative by leveraging the principles of quantum computing to encode and process information in more compact and expressive forms. In this work, we investigate the use of a stacked QAE architecture for cyberattack detection in smart grid environments, aiming to capture both spatial and temporal patterns through quantum-enhanced learning. Preliminary results show that even with minimal input dimensionality, the quantum model performs competitively with classical baselines, highlighting the potential of QAEs as a foundation for scalable, nextgeneration anomaly detection frameworks in critical infrastructure systems.