Enregistré dans:
| Auteur principal: | |
|---|---|
| Format: | Recurso digital |
| Langue: | |
| Publié: |
Zenodo
2025
|
| Accès en ligne: | https://doi.org/10.5281/zenodo.15615585 |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
Table des matières:
- <p>The increased integration of distributed energy resources (DERs) in Active Distribution Systems (ADS)<br>enhances grid flexibility but also introduces new cybersecurity vulnerabilities. Cyber-attacks on such systems<br>can disrupt grid operations, leading to energy instability or blackouts. This paper proposes an adaptive<br>hierarchical framework for cyber-attack detection and localization in ADS. The proposed system combines<br>deep learning with graph-based spectral clustering and statistical impact scoring to identify and locate cyber<br>intrusions in near-real-time. Detection is achieved using a sequential deep neural network trained on electrical<br>signal anomalies, while localization is carried out in two stages—first, using spectral clustering for regional<br>detection, and then using waveform-based metrics for precise node identification. The system’s performance<br>is validated using simulation data from IEEE 37-node feeders with injected false data and coordinated attack<br>scenarios. Results demonstrate superior detection accuracy, localization precision, and scalability compared<br>to conventional intrusion detection systems. This work contributes toward building a resilient smart grid by<br>enabling real-time situational awareness and rapid response</p>