Performance Analysis of VMD to Decompose, Detrend and Denoise Power System Signals

Variational Mode Decomposition (VMD) has gained significant attention as an effective tool for signal processing, particularly in the fields of biomedical and speech processing. This paper explores the application of VMD to decompose complex power system signals which are non-stationary and nonlinear. Standard Empirical Mode Decomposition (EMD) and its variants often encounter challenges like mode mixing, boundary problems, and parameter dependency on noise levels, which may adversely affect the accuracy and reliability of the decomposition results. Since VMD effectively addresses these challenges by providing a more robust framework for decomposition, the resulting Intrinsic Mode Functions (IMFs) have been successfully used for mode estimation, detrending and denoising of power system signals. While denoising, to automate the process of identifying noisy IMFs reliably, Noise Identification Indices (NIIs) have been used. This study employs datasets from 3-machine, 9-bus power system and real-world ISO New England (ISO-NE) power system signals to demonstrate the efficacy and applicability of VMD in practical scenarios. These findings show up the potential application of VMD for analysing power system signals to advance signal processing techniques across various fields.