基于包络分析的齿轮故障监测及基于人工神经网络的齿轮故障分类-gear fault monitoring based on envelope analysis and gear fault classification based on artificial neural network.docxVIP

ABSTRACTGearshasbeenwidely usedasintegralpartsformanyequipmentsand machinesin variousindustries,Therapidgrowthanddevelopmentofmoderntechnologyhascaused the equipments to become more sophisticated and therefore imposing a very challenging taskinmonitoringtheiroperatingconditions.Beingpartofthesecomplexmechanical systemsgearboxshouldbemonitoredcloselyandeffectivelysothatanydefectcanbe detected andtimely corrected initsearly stages.Theenvelopeanalysishaslongbeenwidelyusedforbearingsandgearfault diagnosisbasicallyinvolvingtheuseofresiduesignalobtainedaftersynchronous averagingbeforebandpassfilteringaroundthesignificantgearmeshingcomponents. Howeverinsomeapplicationsthesynchronousaveragingofgearvibrationsignalcan hardlybeachievedandthereforethereneedtobeothertechniquetofacilitateenvelope analysisforgearfaultdiagnostic.Thisstudyisthereforeaimedatsolvingtheabove mentionedproblembyintroducingademodulationtechnique,whichinvolvebandpass filteringaroundtheexitedstructuralresonanceawayfromanysignificantgearmeshing harmonics,todetectandclassifyingeachofthetwotypesoflocalgearfaults(gear toothcrackandspall)withdifferentseveritylevels.Itisknownthatalocalfaulton gear toothwillgeneratesalow-levelimpulseeverytimethefaultcontactsanothergeartooth surfaceduringmeshingcausingamodulationeffectonthegearvibrationsignaland possiblyexcitingthestructuralresonances.Theselow-energyimpactsarealways overwhelmedbythehighenergyvibrationsignalfromothersourcesinthegearbox,so themainideaistodevelopasuitabletechniquewhichcanextractthesedefectinduced impulsesfromwhichthepresenceandseverityofthefaultcanberevealed.The envelope analysis proposed here wasaccomplished by;① Observingthepresenceofdefectinducedresonancesusingspectrumoftheoriginalvibrationsignalcollectedfromthegearboxandthenselectingasuitable demodulationbandfromwhichtheimpulsecausedbyalocalgearfaultsimpactscanbe extracted and used for diagnostic purpose.② Choosingthecenterfrequencyforthebandpassfilteringaroundtheexcitedstructuralresonanceawayfrom