Research Of Rolling Bearing Element Fault Diagnosis Based On Vibration Technology

Rolling bearing element is widely used in rotating machinery, its state hasinfluence on the safety of rotating machinery directly. Then the diagnosis of rollingbearing element with early faults is quite important in engineering field. However,early faults are usually very weak and easily being concealed by strong backgroundnoise. Accordingly, the characteristic extraction of fault signal through digital signalprocessing method is the core of rolling bearing element diagnosis.This paper analyzing the vibration mechanism of rolling bearing element, on thebasis of it a dynamic model was proposed for rolling bearing element with singlelocalized fault. In the model the deformation of race and the geometry character offault were considered, finally the dynamic equation of outer race was get. Numericalanalysis shows the model of raceway fault could simulate the changing rule of impactforce while different rolling element overpass the fault, meanwhile the fault model ofrolling element shows that the impact force caused by fault rolling element isconnected with its angle. By compare with experiment data the model was proved tobe reasonable. It is very valuable for state monitoring and fault diagnosis.As kurtosis has sensitivity to singular signal, it could be used to detect the systemabnormality, however, kurtosis as a global index could not detect the transientinformation of the signal, emergence and development of Spectral Kurtosis (SK) isnecessary. Nowadays SK is a research hotspot in extraction of fault character, it isused in fault diagnosis of rolling bearing element, gear and power system. With thedevelopment of research, some problems of SK in engineering application werediscovered. Firstly, early fault signal of rolling bearing element is usually concealedby strong background noise that could not be detected by SK directly. Secondly theconstruction of Kurtogram is influenced by the sampling points and the number offilter was limited. Hence this paper proposes a new diagnosis method by combineQuantum Genetic Algorithm (QGA) with SK. As SK is sensitive to transientinformation and QGA has ideal global optimization ability, the diagnosis methodcould adaptively search the parameter of filter for envelope analysis. Then computethe time domain envelope and envelope spectrum, the fault character of rollingbearing element could be detected by analyzing envelope spectrum. The results of numerical simulation and experiment both shows that the method is correct.