A vibration signal decomposition method for time-varying structures using empirical multi-synchroextracting decomposition

Abstract

To improve the mode decomposition capacity of the empirical Fourier transform for time-varying structures, an empirical multi-synchroextracting decomposition method has been proposed and applied to mode analysis of time-varying structures. The multi-synchroextracting transform is introduced to obtain the time–frequency coefficients and the time–frequency spectra of response signals. Then, the time–frequency energy of an arbitrary frequency line is obtained by summing the time–frequency coefficients along the entire time history. Due to the total time–frequency energy within the characteristic frequency band is larger than the energy within an adjacent region outside the frequency band, an energy segmentation operator is constructed to determine the energy spectrum boundaries for each mono-component signal. Once the boundaries are found, a zero-phase filter bank and the Fourier transform are used to accomplish frequency spectrum segmentation of the mono-component signal. Finally, each mono-component signal is reconstructed by employing the inverse Fourier transform to each segment, which realizes the mode decomposition of a time-varying structure. The feasibility of the proposed method has been verified against a numerical 2DOF mass-spring-damper system, a numerical three-story frame structure and an experimental twelve-story reinforced concrete frame structure. The analysis results show that the proposed method provides better precision than the empirical Fourier transform in the aspect of time-varying mode decomposition. Moreover, the proposed method has higher decomposition accuracy in the presence of high interference between adjacent frequency bands.