View in EDS

Enhanced Detection of False Data Injection Attacks Using Hybrid Clustering‐Classification for Various Penetration and Distribution Levels of Renewables.

Saved in:
Bibliographic Details
Title: Enhanced Detection of False Data Injection Attacks Using Hybrid Clustering‐Classification for Various Penetration and Distribution Levels of Renewables.
Authors: Pirhadi, Farhad1 (AUTHOR), Seifi, Hossein1 (AUTHOR) seifi_ho@modares.ac.ir, Delkhosh, Hamed1 (AUTHOR)
Source: IET Renewable Power Generation (Wiley-Blackwell). Jan2025, Vol. 19 Issue 1, p1-20. 20p.
Subject Terms: *Renewable energy sources, Anomaly detection (Computer security), Electric power system security measures, Feature selection, Cyberterrorism, Clustering algorithms, Machine learning, Classification algorithms
Abstract: State estimation (SE) is a crucial tool for the secure operation of transmission systems, which are susceptible to false data injection attacks (FDIAs) that can bypass the bad data detection mechanism. The increasing penetration and distribution of renewable energy sources (RESs) reduce the predictability of normal grid behaviour, thereby weakening the performance of conventional detection methods by providing the opportunity for stealthy attacks. This paper presents a new FDIA detection method based on hybrid machine learning (HML), leveraging soft and hard clustering before the classification‐based anomaly detection. It effectively distinguishes between attack and normal samples in grids with various types (wind and solar), distributions, and penetration levels of RESs. The suggested feature engineering enables clustering, using a combination of fuzzy C‐means (FCM) and K‐means, to better organise the database so that a desirable performance can be achieved by the random forest (RF)‐based classifier. The model is comprehensively tested on the IEEE 14‐bus system under 22 RES scenarios, showing robust accuracy across diverse grid conditions. Additionally, the scalability of the method is validated on the IEEE 118‐bus system. The method outperforms recent approaches with average detection accuracies of 99.66% and 99.04% on the small and large systems, respectively. [ABSTRACT FROM AUTHOR]
Copyright of IET Renewable Power Generation (Wiley-Blackwell) is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: GreenFILE
Description
Abstract:State estimation (SE) is a crucial tool for the secure operation of transmission systems, which are susceptible to false data injection attacks (FDIAs) that can bypass the bad data detection mechanism. The increasing penetration and distribution of renewable energy sources (RESs) reduce the predictability of normal grid behaviour, thereby weakening the performance of conventional detection methods by providing the opportunity for stealthy attacks. This paper presents a new FDIA detection method based on hybrid machine learning (HML), leveraging soft and hard clustering before the classification‐based anomaly detection. It effectively distinguishes between attack and normal samples in grids with various types (wind and solar), distributions, and penetration levels of RESs. The suggested feature engineering enables clustering, using a combination of fuzzy C‐means (FCM) and K‐means, to better organise the database so that a desirable performance can be achieved by the random forest (RF)‐based classifier. The model is comprehensively tested on the IEEE 14‐bus system under 22 RES scenarios, showing robust accuracy across diverse grid conditions. Additionally, the scalability of the method is validated on the IEEE 118‐bus system. The method outperforms recent approaches with average detection accuracies of 99.66% and 99.04% on the small and large systems, respectively. [ABSTRACT FROM AUTHOR]
ISSN:17521416
DOI:10.1049/rpg2.70157