Current-, Force-, and Vibration-based Techniques for Induction Motor Condition Monitoring

The aim of this research was to discover the best indicators of induction motor faults, as well as suitable techniques for monitoring the condition of induction motors. Numerical magnetic field analysis was used with the objective of generating reliable virtual data to be analysed with modern signal processing and soft-computing techniques. In the first part of the research, a fuzzy system, based on the amplitudes of the motor current, was implemented for online detection of stator faults. Later on, from the simulation studies and using support vector machine (SVM), the electromagnetic force was shown to be the most reliable indicator of motor faults. Discrete wavelet transform (DWT) was applied to the stator current during the start-up transient, showing how the evolution of some frequency components allows the identification and discrimination of induction motor faults. Predictive filtering was applied to separate the harmonic components from the main current signal. The second part of the research was devoted to the development of a mechanical model to study the effects of electromagnetic force on the vibration pattern when the motor is working under fault conditions. The third part of this work, following the indications given by the second part, is concerned with a method that allows the prediction of the effect of the electromechanical faults in the force distribution and vibration pattern of the induction machines. The FEM computations show the existence of low-frequency and low-order force distributions acting on the stator of the electrical machine when it is working under an electrical fault. It is shown that these force components are able to produce forced vibration in the stator of the machine. This is corroborated by vibration measurements. These low-frequency components could constitute the primary indicator in a condition monitoring system. During the research, extensive measurements of current, flux and vibration were carried out in order to supply data for the research group. Various intentional faults, such as broken rotor bars, broken end ring, inter-turn short circuit, bearing and eccentricity failures, were created. A real dynamic eccentricity was also created. Moreover, different supply sources were used. The measurements supported the analytical and numerical results. Technology (TKK). It was part of a project concerning the diagnosis of faults in electrical machines. Professor Antero Arkkio acted as my supervisor during the research period. I gratefully acknowledge his support and thank him for his guidance. I thank Dr. Anouar Belahcen for his valuable suggestions. I …