Spectrum equalization method based on wavelet denoising algorithm and self-adjusting iteration

Abstract

Spectrum equalization is a key problem of random vibration control. A wavelet spectrum equalization method is proposed in this paper. With multi-resolution transform method, the spectrum is decomposed into fractional scaling signals, after equalization and denoising, they are reconstructed for spectrum estimation. An improved wavelet transform algorithm with FFT and IFFT is introduced to remove the frequency redundance during the decomposition and reconstruction. In order to resolve the contradiction between convergence speed and accuracy during equalization, a self-adjusting iteration algorithm with variable forgetting factor is proposed, the convergence analysis is also carried out. Simulation results show that comparing to traditional WOSA method, the proposed method can achieve better equalization performance, also provides a balance between spectrum oscillation suppression and spectrum discontinuity protection.