Research On Local Oscillatory-characteristic Decomposition Method And Its Application In Fault Diagnosis Of Power Machinery

At present,the research on power machinery condition monitoring has extremely significance with the rapid development of the power machinery.However,most of the vibration signals of the fault machinery is non-stationary and non-linear of time-varying signals.Due to the fact that characteristics of the machinery vibration signals,the new method of Local Oscillatory-characteristic Decomposition(LOD) is applied to fault diagnosis of machinery.In addition,the LOD and analysis of the Hilbert envelope spectrum method were combined for the diagnosis of bearing and gearing fault,and some of LOD theory problem is resolved.The major research contents are as follows:1.On the basis of introducing the algorithm of LOD,this paper makes deeply research for LOD's boundary effect and mode mixing phenomenon and filter bank structure.Firstly,in view of problem of the boundary effect in LOD,a self-adaptive waveform matching extending method is recommended to extend the simulated signal,and the result indicated that the boundary effect in LOD can be resolved effectively.Secondly,based on the statistical characteristics of white noise,the filter bank structure of LOD were obtained by decomposition result of the white noise.Finally,in order to resolve the mode mixing phenomenon in LOD,an improved noise-assisted method based on LOD is proposed.The processed results from experimental and simulated signal indicate this method can resolve the mode mixing phenomenon of LOD effectively.2.The fault characteristic information from rolling bearing and gear in power machinery system that is stored in non-stationary and non-linear,modulation of the vibration signal.In view of this characteristic,based on LOD and analysis of the Hilbert envelope spectrum is proposed and this method used to diagnose the machinery faults.By analyzing gear and rolling bearing experimental signal and actual signal under the fault condition,the processed result indicate that this method can extract information of fault characteristic effectively.