The Application Research Of Adaptive Mathematical Morphology In The Bearing Fault Diagnosis

Bearings are widely used in mechanical systems as important parts of machineryequipment. It will affect the mechanical operation when the bearing is failure. Bearingfailures may lead to mechanical system paralysis, even result in the enterprisecasualties and bring huge economic losses. Statistics show that mechanical failurecaused by bearing failure account for a large proportion of the overall mechanicalfailure. Therefore, it is important to monitor the rolling bearing early fault, find outthe fault occurring point and predict failures development.To extract the fault characteristic information and identify the bearing fault typeand injury point from pending signals is the core and key of the bearing faultdiagnosis. This article study the types and characteristics of bearing failure.Traditional morphological has some issues such as structural elements selection isdifficult and noise reduction effect is not ideal. Generalized morphology differencefiltering algorithm and adaptive tensor morphological algorithms are proposed. Themain contents are as follows:i) The generalized morphological difference filtering of bearing fault featureextraction algorithm is proposed. According to local feature information of signal, theoptimal scale of structural element is selected and generalized differencemorphological filter is built. This new filter can better extract fault features. The resultof utilizing the filter to process the simulation signal and bearing fault simulationexperiment signal shows that the method is effective.ii) To improve the effect of morphological noise reduction and morphologicalfeature extraction, adaptive tensor morphology filer algorithm is proposed based ondifference generalized morphological filtering. A new method builds tensor ellipsestructural element based on the local characteristics information of signal. It canreplace the traditional linear structural elements and disc structure elements. Utilizingtensor morphology filter denoises and extracts fault features of bearing fault signal.The method has been well used in bearing failure.iii) Comparison of three methods the pros and cons in bearing fault featureextraction, a comprehensive analysis of the results show that the effect of adaptivetensor morphology extracts bearing inner and outer ring fault feature is optimal.Generalized morphological difference filter is the second choice. Traditionalmorphology is last choice. However, traditional morphology is the most suitable forbearing rolling element fault extraction. The second is adaptive tensor morphology and generalized morphology.