A Signal Envelope Extraction Method Based on Sparse Restoration

Abstract

The invention discloses a signal envelope extraction method based on sparse restoration, which firstly finds out all maximum value points and all minimum value points in the signal of the envelope line to be extracted, and correspondingly forms the maximum value point vector and The minimum value point vector; then construct a DCT basis, extract row numbers from the DCT basis and each row element consistent with the subscript of each maximum point to form a matrix, and extract the row number and each extreme point from the DCT basis Each row of elements with the same subscript of the small value point forms a matrix; then the maximum value point vector is used as the observation vector, and the corresponding matrix is ​​used as the perception matrix to obtain the upper envelope, and the minimum value point vector is used as the observation vector, corresponding to The matrix is ​​used as a perception matrix to obtain the lower envelope; finally, according to the respective smoothness of the upper envelope and the lower envelope, combined with the change factor of the DCT-based bandwidth to obtain the best upper envelope and the best lower envelope ; The advantage is that it can not only effectively improve the accuracy of the envelope, but also effectively suppress the endpoint effect.

Description

technical field [0001] The invention relates to a signal processing technology, in particular to a signal envelope extraction method based on sparse restoration. Background technique [0002] The core of the envelope analysis method is to demodulate the low frequency fault information modulated in the middle and high frequency bands to the low frequency for analysis and processing, so as to extract the fault information. In the process of mechanical fault diagnosis, due to the influence of the interference signal and noise of the rotating machinery, the signal-to-noise ratio is improved. Therefore, in the fault diagnosis, especially in the fault diagnosis of gearboxes and rolling bearings, the envelope analysis method has advantages that other fault detection methods cannot The role of substitution is currently the most effective method for diagnosing bearing and gear failures. [0003] In signal processing, currently commonly used envelope analysis methods include: Hilbert...

AI Technical Summary

Problems solved by technology

The method of extracting the signal envelope based on the Hilbert transform can effectively extract the modulation frequency and has certain noise resistance, but as the signal-to-noise ratio becomes smaller, the envelope error obtained by the Hilbert transform will gradually increase, resulting in the extracted The signal envelope is not smooth, which affects the accuracy of signal envelope extraction

However, the signal envelope extracted by the method based on cubic spline interpolation to extract the signal envelope not only has a good smoothness, but also can improve the accuracy of the signal envelope extraction when the nodes are gradually encrypted, but the sample The strip interpolation function requires the first and second derivatives of the data at both ends of the data vector, but the required information at the endpoints cannot be obtained from the data curve, so the envelope will swing greatly at the endpoints, forming a very difficult problem of endpoint effects

Images