Linear Narrowband Filtering Method Based on Resonance Enhanced Singular Value Decomposition

Abstract

The invention discloses a linear narrowband filtering method based on resonance enhancement singular value decomposition. According to the method, firstly a measured signal is reasonably intercepted according to an analysis signal length, then the linear narrowband filtering is carried out, the linear narrowband filtering comprises a step of adding a sinusoidal excitation signal whose frequency is equal to a band-pass filter center frequency and whose amplitude is significantly larger than other components in the analysis signal into the signal, constructing a Hankel matrix, carrying out singular value decomposition and using a first singular value to reconstruct a component signal, and subtracting the excitation signal from the reconstructed signal, finally sections of signals are combined, and the precise extraction of a specific frequency signal amplitude and phase information under a strong interference background is realized. According to the method, a frequency domain disorder problem of traditional singular value decomposition is solved.

Description

Technical field: [0001] The invention belongs to the field of signal processing, and relates to a method for accurately extracting the amplitude and phase information of a specific frequency signal in a measured signal through linear narrow-band filtering based on the resonance-enhanced singular value decomposition principle. Background technique: [0002] Singular value decomposition is introduced into signal processing as a theoretically complete mathematical method for data compression, noise reduction, feature extraction, weak signal separation and filter design, etc., and combined with wavelet analysis, EMD and other processing methods in engineering Further improve the signal processing effect in practice. The traditional singular value decomposition has three basic characteristics: linear decomposition, frequency-domain disorder of reconstruction components and band-pass filtering. In the background of complex engineering noise, when there is no obvious amplitude adv...

AI Technical Summary

Problems solved by technology

In the background of complex engineering noise, when there is no obvious amplitude advantage between the eigenfrequency signal and other frequency signals, the traditional singular value decomposition may decompose the eigenfrequency signal into several singular values, so that a single singular value reconstruction is used to extract There are large uncertainties in the method of eigencomponents

For example, the vibration and noise signals of aero-engines contain high-energy airflow and broad-spectrum combustion vibrations, which make the noise reduction processing and feature extraction results of singular value decomposition reconstruction unsatisfactory.

Images