An inverse decaying frequency modulation EMD method for closely spaced modal parameter identification in high precision with Laplace wavelet correlation filtering

Modal parameter identification is a useful tool to reveal natural vibration characteristics in dynamic design and analysis. Due to the influence of closely spaced modes with close frequencies and large dampings, identification accuracy using traditional Hilbert-Huang transform is always unsatisfactory, since the deadline requirement of neighboring intrinsic mode functions cannot be satisfied when the frequency ratio (high frequency versus low frequency) of two adjacent modes is less than two. Therefore, an inverse decaying frequency modulation empirical mode decomposition method is proposed, in order to separate multiple closely spaced modes in high precision, which not only adjusts the frequency ratio by frequency modulation, but also decreases damping by inverse decaying simultaneously. Next, the modal parameters of each separated single mode are picked up by means of Laplace wavelet correlation filtering. The effectiveness of the proposed method is validated by simulated signals in noise and noise-free conditions, and compared with other methods, showing the advantages of high-precision, closely spaced mode identification. In addition, the proposed method is applied to the assembly tightness identification of bolts in an aero-engine rotor, which achieves successful results.