Skip to main navigation Skip to search Skip to main content

A cyber-resilient multi-agent protection scheme for power distribution systems

  • Mohamed M. Elgamal*
  • , Bishoy E. Sedhom
  • , Abdelfattah A. Eladl
  • , V. Oboskalov
  • , Akram Elmitwally
  • , Juan C. Vasquez*
  • , Amir Abdel Menaem
  • *Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Modern distribution networks face significant protection challenges, as conventional relays and existing multi-agent system-based relaying schemes struggle with reliable directional discrimination, high-impedance faults, busbar faults, and cyber vulnerabilities—particularly when voltage measurements are unavailable or costly. This paper proposes a fully distributed, cyber-resilient multi-agent protection scheme for power distribution systems using current-only directional overcurrent relays (DOCRs). Fault direction is determined solely by local current measurements, eliminating voltage transformers and enabling cost-effective and low-complexity implementation. Relay agents exchange only 4-digit binary messages with neighboring agents during fault suspicion, resulting in lightweight communication suitable for legacy infrastructure. The proposed scheme accurately detects both close-in line faults and busbar faults, reliably classifies fault types, and maintains robust performance across a wide range of fault resistances, including high-impedance scenarios. To ensure cyber resilience, each relay is equipped with an anomaly detection unit (ADU) that combines Principal Component Analysis Algorithm (PCAA) and Local Outlier Factor Algorithm (LOFA), enabling precise real-time detection and mitigation of false tripping cyberattacks. Under ideal measurement conditions (noise-free), the ADU achieves 100% classification accuracy, while requiring only 15% of the Multilayer Perceptron (MLP) model training time and 80% of the Isolation Forest Algorithm (IFA) model’s training time, with significantly faster real-time classification. Under degraded or noisy measurement conditions (35 dB SNR), the ADU maintains 98.5% accuracy, outperforming the MLP (96.8%) and IFA (95%) models. Extensive co-simulation integrating power system dynamics and multi-agent system logic validates fast, selective fault clearing across diverse fault types, resistances, topologies, and cyberattacks. The proposed scheme offers a practical, cost-effective, and inherently secure solution ready for real-world deployment in modern distribution networks and microgrids.

Original languageEnglish
Article number111654
JournalInternational Journal of Electrical Power and Energy Systems
Volume176
DOIs
StatePublished - Mar 2026
Externally publishedYes

Bibliographical note

Publisher Copyright:
© 2026 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license. http://creativecommons.org/licenses/by-nc-nd/4.0/

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

  1. SDG 9 - Industry, Innovation, and Infrastructure
    SDG 9 Industry, Innovation, and Infrastructure

Keywords

  • Cybersecurity
  • Fault identification
  • Multi-agent system
  • Overcurrent relay
  • Protection

ASJC Scopus subject areas

  • Energy Engineering and Power Technology
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'A cyber-resilient multi-agent protection scheme for power distribution systems'. Together they form a unique fingerprint.

Cite this