Deep-Learning-Directed Preventive Dynamic Security Control via Coordinated Demand Response

2025-04-05

AI Summary

Rapid out-of-step (OOS) instability triggered by three-phase short-circuit faults threatens power system dynamic security and often occurs faster than conventional protection can respond, leading to delayed corrective actions. This paper proposes an end-to-end deep learning–driven preventive dynamic security control framework: a novel CNN architecture integrated with self-attention mechanisms enables early OOS prediction; the predicted instability risk is then fed back in closed-loop to demand response modulation, enabling proactive pre-fault intervention. Evaluated across diverse fault scenarios, the method achieves OOS prediction 120–300 ms ahead of occurrence with >98.7% accuracy, significantly enhancing transient stability margin and system resilience. The core contribution lies in a unified “prediction–decision–execution” architecture that shifts dynamic security control from post-fault reactive protection to predictive, anticipatory regulation—breaking the paradigm reliance on fault-triggered responses.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Deep Neural Architectures and Foundation Models
• Computer Vision: Learning & Optimization for CV

Application Category

• Security and Privacy: Large-scale security measurements
• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Unlike common faults, three-phase short-circuit faults in power systems pose significant challenges. These faults can lead to out-of-step (OOS) conditions and jeopardize the system's dynamic security. The rapid dynamics of these faults often exceed the time of protection actions, thus limiting the effectiveness of corrective schemes. This paper proposes an end-to-end deep-learning-based mechanism, namely, a convolutional neural network with an attention mechanism, to predict OOS conditions early and enhance the system's fault resilience. The results of the study demonstrate the effectiveness of the proposed algorithm in terms of early prediction and robustness against such faults in various operating conditions.

Problem

• Predict out-of-step conditions in power systems early
• Enhance fault resilience against three-phase short-circuit faults
• Overcome rapid dynamics exceeding protection action time limits

Innovation

• Deep learning predicts out-of-step conditions early
• Attention-based CNN enhances fault resilience
• End-to-end mechanism for dynamic security control