Research On Fault Diagnosis Method For Rotating Machinery Based On Synchrosqueezing Wavelet Transform

With the improvement of industrial development level,rotating machinery has popularized in daily industrial production.There will occur economic loss or even personal safety damage with the enlarged fault probability,in an increasingly complex working environment.The development of fault diagnosis research work on rotating machinery is of great significance to reduce economic losses and avoiding casualties.However,it is of considerable difficulty to implement the rapid and effective extraction of parameterized time-frequency features because of the limitation between time and frequency resolutions,the interference by cross-terms and the influence by additive noises.The introduced time-frequency analysis method Synchrosqueezing Wavelet Transform(SWT)is to synchrosqueeze the parameters of continuous wavelet transform in frequency direction,which can acquire a time-frequency spectra with higher time-frequency resolution.Therefore,this paper carried out in-depth research work on the methods for fault diagnosis method of rotating machinery based on SWT.Firstly,this paper discussed the general situation of development of fault diagnosis research on rotating machinery,analyzed the mechanism of several common faults in rotating machinery and characteristics of the corresponding vibration signals,and introduced the basic theory and algorithms of traditional time-frequency analysis methods and their application in fault diagnosis.The traditional feature extraction methods(such as time-domain feature parameters,energy moments,etc.)and pattern recognition methods(such as probabilistic neural networks,least squares support vector machines,etc.)were studied.On this basis,the application of the pattern classification method based on quantized features in failure mode classification of rotating machinery was studied through the fault experimental data,and its effectiveness was verified.Secondly,the basic principles and algorithms of SWT was introduced in this paper and compared with several traditional time-frequency analysis methods through the analysis and processing of simulation and actual fault experimental signals.It was verified that the time-frequency analysis method based on SWT can obtain time-frequency spectra with higher time-frequency resolution and more prominent time-frequency characteristic performance,which further demonstrated its superior performance in qualitative fault diagnosis.Then,the image matching method was studied,and a fault diagnosis method based on time-frequency feature image matching was proposed,and compared its classification effect with the pattern classification method based on quantitative features.The results showed that the proposed method can identify the running state of the device more effectively and possessed more obvious advantages in fault diagnosis.In addition,SWT is more prominent from the perspective of combination of qualitative diagnosis and quantitative diagnosis.Finally,this article studied the specific implementation method of MATLAB GUI,and a set of multi-feature matching(classification)system for rotating machinery failure mode that integrated time-frequency analysis,pattern matching and time domain analysis,frequency domain analysis and other methods was designed and developed.The system is simple and practical,and can realize automatic classification of rotating machinery faults.