Research On Rotating Machinery Fault Diagnosis Based On Vibration Signal Processing

With the deepening of industrialization,production and use of a variety of equipment become more and more.However,during the start-up and operation of mechanical equipment,there will inevitably be wear,fracture and other unexpected situations.Therefore,mechanical equipment should be promptly and accurately diagnosed after the failure,so as to avoid greater losses.So the fault diagnosis of rotating machinery is discussed in this paper.The main research work is as follows.Aiming at the vibration signal's measurement and acquisition,the vibration signal detection and analysis system is designed.Hardware system is built.The signal acquisition software is developed by using C# programming language and The vibration signal is stored in the computer.Then the Matlab software is used to realize the reading of vibration signal by GUI programming.The fault diagnosis function is realized by using the common signal processing method.In addition,the system can be extended to a variety of fault diagnosis methods.For the rolling bearing's inner fault,blind source separation based on Gabor transform is used to separate the signals from different sensors.Then EMD method is used to decompose the separated signal.Finally the zoom in the local Hilbert envelope spectrum analysis is used to the decomposed signal.More obvious fault characteristics can be obtained than the zoom in the local Hilbert envelope spectrum analysis to the vibration signal solely.According to rolling bearing's inner?outer and roller fault,VMD method is used to decompose the vibration signal.Then the zooming in on local Hilbert envelope amplitude spectrum analysis is used to the decomposed signal.The results show that,the fault characteristics of the bearing can be more obvious compared with the zooming in on local Hilbert envelope amplitude spectrum analysis and the zooming in on local Hilbert envelope amplitude spectrum analysis after EMD decomposition.The validity of the method was verified.A method of status recognition for bearing fault is proposed,which combines thehigher order cumulant and support vector machine(SVM).The higher order cumulants are extracted as the eigenvectors from vibration signal of known fault status.Then the SVM is trained by the eigenvectors.And the SVM is used to validate by the eigenvectors of higher order cumulants which are extracted from the other vibration signal of known fault status.The test results show that the method which combines the higher order cumulant and SVM can obtain a high recognition rate.