Method for fault diagnosis of rotating machinery shaft crack

Abstract

The present invention discloses a method for fault diagnosis of a rotating machinery shaft crack. The method comprises the following steps: 1. dispersing a vibrating signal of a sample shaft, and pre-processing the vibrating signal to remove a data magnitude error; 2. constructing a wavelet basis function based on a vibration pulse model and initiating the function, and performing continuous wavelet transform on the vibrating signal by using the model; 3. calculating Shannon entropy of first n different wavelet coefficients after the vibrating signal is decomposed, to form a fault eigenvector; 4. performing fault classification on the fault eigenvector using a support vector machine; and 5. continuously searching for and updating a wavelet model parameter by using a particle swarm optimization algorithm and a quasi-Newton method based on BFGS, so as to obtain an optimal SVM classification result. According to the method, multi-scale signal analysis is performed by continuous wavelet transform, and fault feature extraction is performed by combining the Shannon entropy, so that the category of different faults can be automatically identified.

Description

technical field [0001] The invention relates to the technical field of fault diagnosis of industrial rotating machinery, in particular to a fault diagnosis method for cracks in rotating shafts of rotating machinery. Background technique [0002] Rotating machinery fault diagnosis has received a lot of attention from industry and academia in recent decades. In the industrial production and manufacturing process, rotating machinery is widely used in the processing and production process, replacing manual labor to create industrial value more efficiently and at low cost. The safe and efficient manufacturing process depends on the good operation of the rotating machinery. During the high-speed and heavy-load operation of the rotating parts, once the rotating shaft parts are damaged, it will directly or indirectly cause the degradation of production performance and cause damage to other rotating parts. More serious failure events may even cause personal injury to front-line prod...

AI Technical Summary

Problems solved by technology

Although this method analyzes in detail the variation trend of fault feature values ​​in the time domain under different fault degrees, it does not use machine learning methods to verify the accuracy of this method for fault prediction and classification

Images