A fault diagnosis method for rotating machinery based on convolution max-min concave penalty algorithm

Abstract

The invention discloses a method for diagnosing a fault in a rotating machine based on a volume positive large-minimum concave penalty algorithm, comprising the steps of: step S1, collecting vibration acceleration response signals and rotational speed signals of the rotating machine, and recording possible fault feature information; step S2, A section of vibration signal is intercepted from the collected data, and the pattern is obtained therefrom with the shift-invariant K-SVD method; step S3, with the obtained pattern as input, the sparse coefficient is solved by the volume positive large minimum concave penalty algorithm, and the sparse coefficient Convolving with the pattern to obtain the reconstructed fault feature; step S4, analyzing the time domain feature and demodulation spectrum feature of the extracted fault feature signal to determine the fault type and complete the fault diagnosis. The present invention proposes to use volume positive maximum minimum concave penalty as the penalty function, which not only improves the amplitude underestimation problem existing in the existing method, but also converts the algorithm solution process from the time domain to the frequency domain, greatly improving the calculation efficiency.

Description

technical field [0001] The invention belongs to the field of fault diagnosis of rotating machinery, and more specifically relates to a fault diagnosis method for rotating machinery based on a rolling positive large minimum concave penalty algorithm. Background technique [0002] Rotating machinery serves in a complex and changeable environment, and failure will lead to serious consequences. Therefore, it is very important to monitor the condition of rotating machinery. The vibration response signal of rotating machinery contains a lot of noise. How to extract the fault characteristics from the noisy vibration signal accurately and quickly is a difficult problem in mechanical fault diagnosis. [0003] Signal processing methods such as Kurtogram, Empirical Mode Decomposition (EMD) and Convolutional Sparse Coding (CSC) have been applied to mechanical fault diagnosis, but there are still shortcomings in the method. For example, the spectral kurtosis method may cause the resonan...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a local type fault diagnosis method for rotating machinery based on the volume positive large minimum concave penalty algorithm for the defect that the existing convolution sparse representation method will underestimate the impact characteristic amplitude

Images