Data-driven structural dynamic characteristics investigation by combining RDT and IVMD method

Abstract

Data-driven structural monitoring results are often mixed with noise, resulting in the inability to identify the dynamic characteristics. In order to develop an efficient approach to correctly estimate the dynamic modal parameters of structures, a method combining random decrement technology (RDT) and improved variational mode decomposition (IVMD) was proposed. The approach focuses on obtaining the free decay response (FDR) with noise directly by RDT from the original monitoring signal rather than pre-denoising for the raw signal and then derives the noiseless FDR using IVMD to get the dynamic parameters with fewer data. First, the steps of the IVMD principle were established based on traditional VMD. Next, the performance of IVMD was verified. The simulated signal test shows that the IVMD addresses the shortcoming of the VMD algorithm and has better advantages in signal decomposition and reconstruction than complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN). Finally, the frequencies and damping ratios of two large-scale civil structures (a super-high-rise building and a suspension bridge) in the field experiments were captured from the signal response data via RDT-IVMD, proving the reliability of the presented method. The results can provide a basis for structural preventive maintenance.