Deep Learning Based Cybersecurity Enhancement Strategy in Microgrids

Emerging microgrids rely on data communication for control and coordinated operation, making them vulnerable to cybersecurity threats, particularly False Data Injection (FDI) attacks. This paper proposes an advanced cybersecurity enhancement strategy that leverages a deep learning based autoencoder solution for real-time anomaly detection. Upon detecting any anomaly, the proposed framework integrates superimposed component analysis to identify manipulated sensor data, effectively distinguishing it from non-FDI anomalies, such as physical faults. This approach is designed with the considerations of using resource-constrained edge devices and ensuring low-latency performance. The proposed method is validated through extensive microgrid simulations that demonstrates high accuracy in both anomaly detection and classification.