Cyber-Resilient Detection of Power Quality Events With NSCT and PCA-SVM

The increasing reliance on smart grids, coupled with the integration of renewable energy and growing cyber-physical interactions, has heightened the vulnerability of power systems to both power quality (PQ) disturbances and cyber-attacks. This paper presents an innovative detection framework that combines Nonsubsampled Contourlet Transform (NSCT) with Principal Component Analysis (PCA) and Support Vector Machine (SVM) classification to accurately detect and classify PQ disturbances under the influence of cyber threats, such as False Data Injection (FDI) and Denial of Service (DoS) attacks. The proposed methodology leverages NSCT’s multiscale decomposition capabilities to extract fine-grained signal features, while PCA optimizes feature selection for enhanced computational efficiency. Comprehensive experiments conducted on synthetic and real-world datasets validate the framework’s effectiveness, demonstrating supe- rior detection accuracy, robustness to noise, and resilience against cyber-attacks. The proposed NSCT-PCA-SVM approach represents a significant step forward in ensuring secure and reliable smart grid operations.