Adaptive Feature Purification of Cavitation Noise Modulation Spectrum

To enhance the stability of modulation spectrum structure, in this study, a novel adaptive feature purification method called variational mode decomposition (VMD) combined with feature energy entropy (FEE) is proposed. First, to decompose the noise interference envelope into simple intrinsic mode functions (IMFs), the VMD algorithm is employed. Subsequently, to compute the weight space of the IMFs, the FEE algorithm is utilized. Finally, using the respective weights, the envelope signal is reconstructed, which leads to a further enhancement in the stability of the modulation spectrum. Both simulation and real data confirm the effectiveness of the VMD-FEE algorithm in feature purification of the modulation spectrum.