BESSAM Besma2024-02-132024-02-132016http://dspace.univ-khenchela.dz:4000/handle/123456789/718The knowledgeofthebrokenrotorbarscharacteristicfrequenciesandamplitudeshasagreatimportance for allrelateddiagnosticmethods.Themonitoringofmotorfaultsrequiresahighresolutionspectrumto separatedifferentfrequencycomponents.TheDiscreteFourierTransform(DFT)hasbeenwidelyusedto achieve theserequirements.However,atlowslipthistechniquecannotgivegoodresults.Asasolution for theseproblems,this paper proposesanefficient techniquebasedonaneuralnetworkapproachand Hilbert transform(HT)forbrokenrotorbardiagnosisininductionmachinesatlowload.TheHilbert transform isusedtoextractthestatorcurrentenvelope(SCE).Twofeaturesareselectedfromthe(SCE) spectrum (theamplitudeandfrequencyoftheharmonic).Thesefeatureswillbeusedasinputforneural network.Theresultsobtainedareastonishinganditiscapabletodetectthecorrectnumberofbroken rotor bars under different load conditions.enArticle