Research On Fault Analysis Method And Diagnosis System Of Rotating Machinery

Mechanical equipment is widely used in modern production life. If any component fails and is not discovered timely, the results can not only lead the mechanical equipment to be damaged completely, but even cause serious consequences of machine and people damage. With the development of intelligent machinery fault diagnosis technology, huge economic loss can be avoided by applying the intelligent technology into the production practice, and the number of accident casualties can be largely reduced. Therefore, the intelligent machinery fault diagnosis technology has certain significance. Mechanical failure mechanism is learned in this paper firstly. Then a fast and effective online fault detection system is achieved combining with modern methods of artificial intelligence technology. The main content of my paper is as follows:Firstly, the related theory of mechanical failure is introduced with the common bearing and gear failure as an example. Research of bearing and gear failure mechanism and failure types is included particularly. And statistical time domain analysis method, frequency spectrum analysis method and time frequency domain analysis theory method during my research are respectively introduced as detail.Secondly, the bearing fault intelligent diagnosis system with Lab V IEW software is completed based on the time domain statistical method and neural network theory. The characteristic values are extracted after collecting the bearing vibration signals under different working states, which is worked as the target sample and training sample of neural network system. Thresh old is set to judge whether the bearing is on its normal state.Thirdly, local decomposition method(LMD) is studied. The disadvantage of LMD algorithm that it is sensitive to high frequency noise, is found through repeated verification of the simulation s ignal and the measured signal. In order to make up for this short coming, an improved LMD algorithm is proposed, based on extracting the extrema of envelope curve. And three different de-noising methods, i.e. the band-pass filter method, wavelet method and lift wavelet method are respectively used to comparing the decomposition results. Experimental results show that this improved algorithm proved to have good reproducibility and validity.Finally, the fault detection system of speed reducer machine based o n the Lab VIEW development platform is designed and applied to industrial field analysis. The system mainly consists of real time data’s display and preservation, characteristic value’s extraction and preservation which is used to establish the different reducer model’s sample library, and the intelligent diagnosis result’s display.