QSTAformer: A Quantum-Enhanced Transformer for Robust Short-Term Voltage Stability Assessment against Adversarial Attacks

2025-11-29

AI Summary

To address the insufficient adversarial robustness of short-term voltage stability assessment (STVSA) models, this paper proposes QSTAformer—a quantum-enhanced Transformer architecture. It integrates parameterized quantum circuits (PQCs) into the Transformer’s attention mechanism to form a quantum-classical hybrid attention module and introduces a customized adversarial training strategy compatible with both white-box and gray-box attack settings. Key contributions include: (1) the first systematic analysis of adversarial vulnerability in quantum machine learning for STVSA; (2) the pioneering paradigm of co-designing quantum circuits and attention mechanisms; and (3) multi-objective PQC architecture optimization balancing expressivity, convergence, and computational efficiency. Evaluated on the IEEE 39-bus system, QSTAformer matches classical models in accuracy, reduces inference complexity by 23%, and improves robustness against FGSM and PGD attacks by over 41%, significantly enhancing security-aware STVSA under malicious perturbations.

Technology Category

• Machine Learning: Quantum Machine Learning
• Computer Vision: Adversarial Attacks & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Security and Privacy: Large-scale security measurements

Abstract

Short-term voltage stability assessment (STVSA) is critical for secure power system operation. While classical machine learning-based methods have demonstrated strong performance, they still face challenges in robustness under adversarial conditions. This paper proposes QSTAformer-a tailored quantum-enhanced Transformer architecture that embeds parameterized quantum circuits (PQCs) into attention mechanisms-for robust and efficient STVSA. A dedicated adversarial training strategy is developed to defend against both white-box and gray-box attacks. Furthermore, diverse PQC architectures are benchmarked to explore trade-offs between expressiveness, convergence, and efficiency. To the best of our knowledge, this is the first work to systematically investigate the adversarial vulnerability of quantum machine learning-based STVSA. Case studies on the IEEE 39-bus system demonstrate that QSTAformer achieves competitive accuracy, reduced complexity, and stronger robustness, underscoring its potential for secure and scalable STVSA under adversarial conditions.

Problem

• Enhances short-term voltage stability assessment robustness
• Defends against adversarial attacks using quantum-enhanced transformers
• Benchmarks quantum circuit designs for efficiency and convergence

Innovation

• Quantum-enhanced Transformer with parameterized quantum circuits
• Adversarial training strategy for white-box and gray-box attacks
• Benchmarking diverse PQC architectures for trade-offs exploration