Quantum Adversarial Machine Learning for Robust Power System Stability Assessment

Quantum Machine Learning (QML) has emerged as a promising paradigm for addressing power system Transient Stability Assessment (TSA) challenges leveraging the unique expressibility and exponential scalability of quantum operators. However, as a data-driven approach, QML-based TSA is susceptible to malicious attacks on the input data, presenting an underexplored vulnerability in adversarial environments. This paper devises a Quantum adversarial machine learning-based TSA (QaTSA) to offer an efficient and robust quantum-enabled, data-driven TSA under adversarial attacks. Our contributions include 1) a Fourier-type decomposition for analyzing the adversarial vulnerability of QML-based TSA; 2) a Robustness-enhanced, High-Expressibility, Low-Depth (ReHELD) quantum circuit to enhance the robustness of QML-based TSA against attacks; and 3) a systematical evaluation of the performance of QaTSA in diverse attack scenarios. Experimental results validate the effectiveness of QaTSA and provide valuable insights into developing attack-robust and noise-tolerant quantum algorithms for various power system applications.