Zhendong yu chongji · 2012 · 31 citations · 0 references
ComputingMechanical EngineeringMachine LearningTime Series AnalysisVibrationDiagnosisFault DetectionMachine VibrationSupport Vector MachinesVibration DiagnosticsVibration MonitoringBearing Fault DiagnosisBearing FaultsBearing VibrationBearingsEmpirical Mode DecompositionEmpirical Mode Decomposition FeaturesEntropy EstimationVibration-based Diagnostics
The nonlinear dynamic parameter sample entropy was used as a feature for roller bearing fault diagnosis.Vibration signals for normal bearings,those with inner race fault,ball one,and outer race one were used for analysis and diagnosis.The sample entropy of the original vibration signal could be analyzed only in one scale,but information about the characteristics of the vibration signal in different scales could give important information about the fault.A sample entropy method based on ensemble empirical mode decomposition(EEMD) was proposed here.Firstly,the original roller bearing vibration signal was decomposed with EEMD and the intrinsic mode functions containing the most information were chosen to calculate the sample entropy to form a feature vector.Then,SVM method was used as a classifier to identify different faults.Thus,the vibration signal could be analyzed in different scales to give more information about fault.Experimental results with real roller bearing data showed that the proposed method is effective.