Open Access

ARTICLE

A Combined Denoising Method of Adaptive VMD and Wavelet Threshold for Gear Health Monitoring

Guangfei Jia^*, Jinqiu Yang, Hanwen Liang

School of Mechanical Engineering, Hebei University of Science and Technology, Shijiazhuang, 050018, China

* Corresponding Author: Guangfei Jia. Email:

(This article belongs to the Special Issue: Sensing Data Based Structural Health Monitoring in Engineering)

Structural Durability & Health Monitoring 2025, 19(4), 1057-1072. https://doi.org/10.32604/sdhm.2025.061805

Received 03 December 2024; Accepted 13 February 2025; Issue published 30 June 2025

Abstract

Considering the noise problem of the acquisition signals from mechanical transmission systems, a novel denoising method is proposed that combines Variational Mode Decomposition (VMD) with wavelet thresholding. The key innovation of this method lies in the optimization of VMD parameters K and using the improved Horned Lizard Optimization Algorithm (IHLOA). An inertia weight parameter is introduced into the random walk strategy of HLOA, and the related formula is improved. The acquisition signal can be adaptively decomposed into some Intrinsic Mode Functions (IMFs), and the high-noise IMFs are identified based on a correlation coefficient-variance method. Further noise reduction is achieved using wavelet thresholding. The proposed method is validated using simulated signals and experimental signals, and simulation results indicate that the proposed method surpasses original VMD, Empirical Mode Decomposition (EMD), and wavelet thresholding in terms of Signal-to-Noise Ratio (SNR) and Root Mean Square Error (RMSE), and experimental results indicate that the proposed method can effectively remove noise in terms of three evaluation metrics. Furthermore, compared with Feature Mode Decomposition (FMD) and Multichannel Singular Spectrum Analysis (MSSA), this method has a better envelope spectrum. This method not only provides a solution for noise reduction in signal processing but also holds significant potential for applications in structural health monitoring and fault diagnosis.

---

Graphic Abstract

---

Keywords

---