The using of data mining and soft computing techniques in fault diagnosis

ABSTRACTMaster ThesisTHE USING OF DATA MINING AND SOFT COMPUTING TECHNIQUES IN FAULTDIAGNOSISİlhan AYDINFirat UniversityGraduate School of Natural and Applied SciencesDepartment of Computer Engineering2006, Page: 121Induction motors are used in a big part of the field of electromechanical energyconversion in industry. Reliability, robustness and cheap of the costs of these type motors hasbeen provided the preferring of them in many applications. But effects such as operationenvironment and humidity cause faults in different parts of these motors. The goal of faultdiagnosis is to detect these occurred faults at an early stage.In this study, faults occurred in stator, rotor and bearing of induction motors havebeen diagnosed via soft computing and data mining techniques. These techniques use easilyacquired signal such as rotor speed and motor?s current for fault diagnosis. Consequently,mathematical dynamics and internal construction of motors are needn?t known. Broken rotorbars, bearing friction, eccentricity and stator winding faults have been diagnosed via softcomputing techniques such as artificial neural networks, fuzzy logic and artificial immunesystems, successfully. Faults in two different type motors have been detected by combiningfuzzy logic and artificial immune systems. Features to diagnose the stator winding and bearingfriction faults have been extracted from stator current and rotor speed by time series data miningmethod.Two type induction motors have been used for fault diagnosis. Simulation data havebeen obtained from single phase induction motor. Experimental data has been transferred thecomputer by using data acquisition card. Three current sensors have been used for measuringthe current data.Key Words: Fault Diagnosis, Fault Detection, Soft Computing Techniques, Data Mining, TimeSeries Data Mining, Induction Motors