Bayesian optimized Hilbert-Huang GenoMap SVM drift classifier for adaptive PQD classification
R Shilpa, P. S. Puttaswamy · e-Prime – Nexus of Electrical Electronic and Intelligent Engineering · 2026
Power Quality Disturbances (PQDs) are deviations from ideal voltage levels that critically impact equipment and system performance, making accurate real-time classification essential for reliable and secure grid operation. Hence, a novel Bayesian Optimized Hilbert-Huang GenoMap SVM Drift Classifier is proposed for adaptive PQD classification. Initially, the existing EEMD model discards critical short-duration transients, and coupled with feature drift from harmonic distortions, causes unstable decision boundaries and misclassifications in real-time PQD detection. Thus, a novel Hilbert-Huang Bayesian Sparse Drift Extractor (HH-BSDE) is proposed to preserve critical transients, compensate for feature drift, and stabilize the classification boundary in real-time PQD classification. Further, in PQD classification, the absence of adaptive feature selection and feature-ranking in hyperplane-based separation leads to suboptimal performance, limits real-time reliability in high-stakes power systems. Thus, a novel GenoMap-Evolved IntelliSVM (GMEI-SVM) method is proposed to adaptively select and rank features, enhancing real-time adaptability, noise resilience, and robust, accurate PQD classification with high feature-ranking fidelity. The outcomes obtained by the proposed model have high accuracy, precision, and recall, and low delay and detection time, and also provide a robust and interpretable solution for dynamic power grid applications, ensuring operational reliability and safeguarding critical electrical infrastructure.