Machine-Learning Driven Load Shedding to Mitigate Instability Attacks in Power Grids

2025-09-30

AI Summary

Increasing instability-inducing cyberattacks on power systems remain inadequately addressed by existing defenses, which lack both effectiveness and real-time response capability. Method: This paper proposes a data-driven intelligent load shedding defense mechanism that integrates supervised machine learning with modified Prony analysis (MPA) to enable high-accuracy detection and millisecond-scale response to oscillatory instability attacks. The approach models the IEEE 14-bus system and employs Achilles Heel Technologies to generate multi-scenario attack data; a lightweight classifier is trained to identify anomalous dynamic patterns and trigger adaptive load shedding. Contribution/Results: Experimental evaluation demonstrates that the mechanism provides early warning of instability 200–500 ms in advance, reduces system instability probability by 76.4%, and significantly enhances grid resilience and robustness against sophisticated cyberattacks.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Security and Privacy: Large-scale security measurements

Abstract

Every year critical infrastructure becomes more complex and we grow to rely on it more and more. With this reliance, it becomes an attractive target for cyberattacks from sophisticated actors, with one of the most attractive targets being the power grid. One class of attacks, instability attacks, is a newer type of attack that has relatively few protections developed. We present a cost effective, data-driven approach to training a supervised machine learning model to retrofit load shedding decision systems in power grids with the capacity to defend against instability attacks. We show a proof of concept on the IEEE 14 Bus System using the Achilles Heel Technologies Power Grid Analyzer, and show through an implementation of modified Prony analysis (MPA) that MPA is a viable method for detecting instability attacks and triggering defense mechanisms.

Problem

• Machine learning detects instability attacks in power grids
• Load shedding systems defend against cyberattacks on infrastructure
• Modified Prony analysis triggers protection mechanisms for grid stability

Innovation

• Machine learning model for load shedding
• Data-driven approach to detect instability attacks
• Modified Prony analysis triggers defense mechanisms