Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Artificial Intelligence based Approach for Identification and Mitigation of Cyber-Attacks in Wide-Area Control of Power Systems

Jishnudeep Kar, Aranya Chakrabortty

TL;DR

This work tackles cyber-attack threats on wide-area damping control in power systems by introducing a unified LSTM-based GAN framework deployed at generating stations. The encoder-decoder generator imputes attacked state entries while the discriminator detects anomalies, enabling simultaneous identification and mitigation of false data injection and DoS attacks without separate detectors or mitigators. A reconstruction loss with a masking scheme and a moving-average defense improve robustness against false alarms, and the method yields a revised control law that uses imputed states for resiliency. Validated on the IEEE 68-bus system, the approach achieves timely attack detection, accurate state imputation, and maintained stability under attack, making it scalable for decentralized deployment in real grids.

Abstract

We propose a generative adversarial network (GAN) based deep learning method that serves the dual role of both identification and mitigation of cyber-attacks in wide-area damping control loops of power systems. Two specific types of attacks considered are false data injection and denial-of-service (DoS). Unlike existing methods, which are either model-based or model-free and yet require two separate learning modules for detection and mitigation leading to longer response times before clearing an attack, our deep learner incorporate both goals within the same integrated framework. A Long Short-Term Memory (LSTM) encoder-decoder based GAN is proposed that captures the temporal dynamics of the power system significantly more accurately than fully-connected GANs, thereby providing better accuracy and faster response for both goals. The method is validated using the IEEE 68-bus power system model.

Artificial Intelligence based Approach for Identification and Mitigation of Cyber-Attacks in Wide-Area Control of Power Systems

TL;DR

This work tackles cyber-attack threats on wide-area damping control in power systems by introducing a unified LSTM-based GAN framework deployed at generating stations. The encoder-decoder generator imputes attacked state entries while the discriminator detects anomalies, enabling simultaneous identification and mitigation of false data injection and DoS attacks without separate detectors or mitigators. A reconstruction loss with a masking scheme and a moving-average defense improve robustness against false alarms, and the method yields a revised control law that uses imputed states for resiliency. Validated on the IEEE 68-bus system, the approach achieves timely attack detection, accurate state imputation, and maintained stability under attack, making it scalable for decentralized deployment in real grids.

Abstract

We propose a generative adversarial network (GAN) based deep learning method that serves the dual role of both identification and mitigation of cyber-attacks in wide-area damping control loops of power systems. Two specific types of attacks considered are false data injection and denial-of-service (DoS). Unlike existing methods, which are either model-based or model-free and yet require two separate learning modules for detection and mitigation leading to longer response times before clearing an attack, our deep learner incorporate both goals within the same integrated framework. A Long Short-Term Memory (LSTM) encoder-decoder based GAN is proposed that captures the temporal dynamics of the power system significantly more accurately than fully-connected GANs, thereby providing better accuracy and faster response for both goals. The method is validated using the IEEE 68-bus power system model.
Paper Structure (12 sections, 8 equations, 9 figures, 2 algorithms)

This paper contains 12 sections, 8 equations, 9 figures, 2 algorithms.

Table of Contents

  1. Introduction
  2. Problem Statement
  3. Proposed GAN Design Framework
  4. 1. Implementation architecture
  5. 2. Redesigned training of the GAN
  6. 3. Usage for online identification and resiliency
  7. Detection and identification of cyber-attack :
  8. Resiliency using GAN based imputation :
  9. Numerical Results
  10. 1. Training of GAN
  11. 2. False data injection and DoS attacks
  12. Attack Mitigation :

Figures (9)

  • Figure 1: Proposed LSTM-GAN framework and associated learning architecture
  • Figure 2: IEEE 68-bus, 16-machine power system model
  • Figure 3: Generator vs discriminator score (training)
  • Figure 4: FDI attack destabilizing the system
  • Figure 5: Attack detection using discriminator output
  • ...and 4 more figures