A Classical-Quantum Transfer Learning Model for Disturbance Detection in Power Systems

Abstract

In power systems, disturbances such as faults, loss of generation, and synchronous motor switching can interrupt the power supply, causing significant damage. Wide Area Monitoring, Protection, and Control (WAMPAC) systems are employed to maintain stable grid operation. With the increasing adoption of artificial intelligence (AI), AI-based methods are being integrated into WAMPAC systems to enhance their operational capabilities. Beyond classical machine learning (ML) approaches, quantum machine learning (QML) offers the potential to advance power systems research by incorporating the principles of quantum mechanics. In this study, we propose a hybrid classical-quantum transfer learning model based on EfficientNetB0 for detecting common disturbances in power systems. We compared its performance with established classical ML models, including artificial neural networks (ANN), convolutional neural networks (CNN), and a conventional transfer learning (TL) model. Our results demonstrate that hybrid classical-quantum models are a viable alternative to classical approaches. However, challenges such as class imbalance can hinder their performance. These issues must be addressed effectively to realize a true quantum advantage for practical applications.