An Intelligent Optimization Algorithm's Evaluation and Comparison with the Evolutionary Algorithm for PD Denoising and Improved Convergence

Abstract

Introduction: Partial discharge (PD) signals measured in high-voltage systems are frequently affected by various forms of noise, which complicates accurate detection and analysis. Traditional denoising methods that rely on single-domain decomposition and basic optimization often fail to sufficiently remove noise while retaining critical PD details.

Objectives: The objective of this study is to design an advanced denoising strategy that effectively suppresses noise while preserving important PD characteristics to improve diagnostic reliability.

Methods: A hybrid denoising framework combining Variational Mode Decomposition (VMD), Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), and Singular Value Decomposition (SVD) is presented. The PD signal is decomposed into intrinsic mode functions (IMFs), and reconstruction is refined using power spectral entropy to discard noise-dominated components. Furthermore, an Improved Zebra Optimization Algorithm (IZOA) is integrated with the Whale Optimization Algorithm (WOA) to ensure an optimal balance between exploration and exploitation. The proposed method is assessed through simulation studies and compared with other optimization-based denoising techniques using multiple evaluation metrics.

Results: The proposed Evalutionary algorithm demonstrates superior noise removal capability, preserves essential PD features, and achieves significantly improved performance metrics compared to traditional denoising approaches.

Conclusions: The VMD–CEEMDAN–SVD hybrid denoising scheme supported by IZOA–IWOA optimization provides an efficient and robust solution for PD signal processing, improving the accuracy and dependability of high-voltage insulation condition monitoring.