Hybrid STSMC Observer and AI-Based Anomaly Detection in Smart Grids with EV Integration
2026 IEEE International Conference on Cybersecurity, Digital Forensics, and AI Applications (ICCSDFAI), İstanbul, Türkiye, 25 Haziran 2026, ss.486-492, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/iccsdfai70505.2026.11647996
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Sayfa Sayıları: ss.486-492
- Gazi Üniversitesi Adresli: Evet
Özet
The increasing penetration of renewable energy sources and electric vehicles (EVs) has significantly increased the operational complexity and uncertainty in modern smart grids, making reliable anomaly detection an important requirement for system monitoring and security. Conventional model-based methods provide physically interpretable results but may show limited performance under highly nonlinear and dynamic conditions, while artificial intelligence (AI)-based approaches offer strong pattern recognition capabilities at the expense of reduced physical transparency. In this study, a hybrid anomaly detection framework is proposed by combining a Super-Twisting Sliding Mode Controller (STSMC)-based observer with an AI-based detection model. The STSMC observer is used to estimate system states and generate residual signals, which are then used as inputs to the AI model instead of raw measurement data. This structure improves detection sensitivity and reduces false alarm rates by enhancing the quality of the input features. The proposed method is evaluated under multiple operating conditions, including EV charging anomalies, renewable generation variations, grid faults, and cyber-attack scenarios. The results show that the hybrid approach provides higher detection accuracy, faster response, and improved robustness compared to conventional model-based and AI-only methods, demonstrating its effectiveness for smart grid applications with high renewable and EV penetration.