Evaluating Signal Processing Methods for Instantaneous Frequency Analysis in Time-Varying Mass Structures

Chinnapat Buachart (Department of Civil Engineering, Faculty of Engineering, Chiang Mai University, Chiang Mai 50200)
Chayanon Hansapinyo (Department of Civil Engineering, Faculty of Engineering, Chiang Mai University, Chiang Mai 50200)
Teewara Suwan (Department of Civil Engineering, Faculty of Engineering, Chiang Mai University, Chiang Mai 50200)
Vanissorn Vimonsatit (School of Engineering, Faculty of Science and Engineering, Macquarie University, Sydney, NSW 2019)
Phung Tu (Flow Without Quake, Perth, Western Australia 6000)
Worathep Sae-Long (Civil Engineering Program, School of Engineering, University of Phayao, Phayao 56000)
Suchart Limkatanyu (Department of Civil and Environmental Engineering, Faculty of Engineering, Prince of Songkla University, Songkhla 90112)
Panatchai Chetchotisak (Department of Civil Engineering, Rajamangala University of Technology Isan, Khon Kaen Campus, Khon Kaen 40000)

Civil Engineering Journal
Vol. 12 No. 4 (2026): April

Time-varying mass (TVM) structures exhibit complex dynamic phenomena but remain insufficiently investigated, particularly in the frequency domain. For example, granular discharge in silos generates vibrations due to rapid mass reduction, leading to nonlinear and non-stationary responses. This study aims to evaluate the capability of three common signal processing algorithms-Empirical Mode Decomposition (EMD), Variational Mode Decomposition (VMD), and Empirical Wavelet Transform (EWT)-for analysing instantaneous frequency variations in TVM structures. The signals are first decomposed into mono-component modes and subsequently analyzed using the Hilbert transform to extract instantaneous frequency. The investigation is conducted in two stages: (i) numerical validation using an artificial nonlinear signal and a time-varying parameter SDOF system with known frequency histories, and (ii) application to experimental acceleration data obtained from sand discharge in a polycarbonate silo under noisy conditions. The findings show that EMD provides the most accurate frequency estimation for clean signals, whereas VMD and EWT demonstrate improved stability for experimental data with significant noise. The study provides a systematic comparison of decomposition-based instantaneous frequency methods in TVM structures and highlights the importance of appropriate method selection for safer and more reliable frequency-domain structural design.