Detection on rotating mechanical asymmetries in small induction machines

The main objective of this research work is to develop a current-based, sensorless scheme for early detection of mechanical faults and asymmetries in induction machines. A scheme that is independent of load variations, uses simple approaches, and minimizes data processing and storage.; Conventional approaches to electric machines condition monitoring and fault detection are reviewed, covering both on-line and off-line methods and focusing on vibration monitoring and stator current monitoring, which are commonly used for detecting mechanical faults in induction machines. Comprehensive testing of various mechanical fault conditions covering different levels of each fault and different types of induction machines and bearings is carried out to show the changes in the machine vibration and stator current as functions of fault severity.; An evaluation of using bearings with faults generated off-line is performed, emphasizing the disadvantages of disassembling and remounting the bearings and demonstrating the effects of these acts on the fault detection schemes. Two new approaches to fail the bearings on-line are experimentally investigated: applying a radial force on the shaft or directly on the bearing, and applying a DC current through the front bearing.; Comprehensive testing of induction motors of different horsepower and different pole-sizes is conducted for rotor unbalance and shaft misalignment detection, and a simple technique for mechanical fault detection in induction machines using stator current is proposed. New ways for individually creating static and dynamic air-gap eccentricities are examined, and a new method for separating arbitrary load torque effects from air-gap eccentricity effects on the stator current is proposed.; The effects of supplying the induction motor from an adjustable speed drive (in an open-loop control) on the detection of rotor unbalances and radial shaft misalignments are investigated. Finally, a new algorithm for detecting mechanical fault conditions in induction motors under any load condition using spectral analysis of the stator current is derived. The proposed algorithm takes into account the abnormalities and the resonance caused by changing the load.