Rubbing Fault Diagnosis Method Based on Greedy Sparse Identification of Wavelet and Harmonic Components

Abstract

The invention discloses a rubbing fault diagnosis method based on greedy and sparse recognition of wavelet and harmonic components, and relates to a mechanical fault diagnosis method. First, the redundant second-generation wavelet packet transform is used to decompose the signal at multiple scales; the amplitude signal is emphasized from the subspace signals of each scale; the Hanning window Fourier transform is added to the selected subspace, and the signal is divided into two parts according to the principle of energy concentration Several harmonic components; the greedy sparse identification algorithm of harmonic parameters is used to iteratively and accurately identify the amplitude, frequency and phase parameters; the noise reduction signal of subspace is constructed by harmonic parameters; demodulation is carried out by Hilbert envelope The instantaneous amplitude and instantaneous frequency information of the subspace can be accurately obtained to diagnose the rubbing fault of the rotor system. It has precise translation invariance and linear phase characteristics. It can avoid the pattern rupture phenomenon in the decomposition process, continuously improve the identification accuracy of parameters through the iterative method, and diagnose the rubbing fault of the rotor system through the periodic mutation of instantaneous amplitude and instantaneous frequency.

Description

technical field [0001] The invention relates to a mechanical fault diagnosis method, in particular to a rubbing fault diagnosis method based on greedy and sparse recognition of wavelet and harmonic components, which reduces the noise of vibration displacement signals to improve the recognition accuracy of instantaneous parameters. Background technique [0002] Rubbing is a common type of failure in the rotor system, which often occurs between the rotor-stator of aero-engines and power machinery. The eddy current sensor can measure the vibration displacement signal of the rotating shaft in a non-contact manner, and is a reliable carrier of the rubbing signal. The fault features produced by early rubbing faults are weak and difficult to be directly identified from the original signal, and advanced signal processing methods are needed to extract them. [0003] The time-frequency analysis method is a common tool for vibration displacement signal analysis. Gong Xiaoyun et al. (...

AI Technical Summary

Problems solved by technology

[0004] Although the current time-frequency analysis method can adaptively select the filter passband according to the characteristics of the rubbing signal, thereby suppressing the noise outside the passband, it still cannot filter out the noise within the passband, thus affecting the instantaneous amplitude and instantaneous frequency. The correct extraction of parameters such as

Images