Shafting Fault Identification Method Based on Wavelet Threshold Noise Reduction and Adaboost

Abstract

The invention provides a shaft system fault identification method based on wavelet threshold de-noising and AdaBoost. According to the method, a dual-tree complex wavelet transform method capable of eliminating frequency aliasing is adopted to extract the features of signals; in the process of decomposition and reconstruction of the signals, an improved wavelet threshold de-noising method is provided to perform de-noising processing on the signals; the energy of the de-noised signals is extracted and adopted as a feature vector; the AdaBoost multi-classification method having a good unbalanceddata classification effect is used in combination; a plurality of simple single-level decision trees are used as AdaBoost weak classifiers; and a strong classifier can be constructed finally to distinguish various shaft system faults. The method of the invention can be realized through programming and has the advantages of low cost, high efficiency and easiness in implementation.

Description

technical field [0001] The invention belongs to the field of fault identification, and in particular relates to a shaft fault identification method of mechanical equipment. Background technique [0002] The most effective way to diagnose the shafting fault of rotating machinery is to analyze the fault through the shafting vibration signal. The industrial production site environment is relatively complex, and the obtained vibration signal contains a lot of noise. The existence of a large amount of noise seriously affects the accuracy of shafting fault identification. How to effectively reduce noise and remove noise has always been a hot research topic. The vibration signal belongs to the non-stationary time signal, and it is often processed by wavelet transform, and the corresponding wavelet denoising method has been widely used. Among them, the common wavelet denoising methods include: using wavelet transform modulus maxima to denoise; based on wavelet transform inter-scale...

AI Technical Summary

Problems solved by technology

[0003] Both hard threshold and soft threshold denoising are somewhat rough in setting. Hard threshold denoising does not process wavelet coefficients greater than the threshold, resulting in insufficient denoising. Soft threshold denoising directly subtracts the threshold when processing wavelet coefficients greater than the threshold, resulting in excessive denoising. dry

At present, some experts and scholars combine hard threshold and soft threshold to comprehensively remove dryness when removing dryness, but the effect is different and unsatisfactory.

Images