FAULT DIAGNOSIS OF MECHANICAL SYSTEMS BASED ON ELECTRICAL SUPPLY CHARACTERISTICS

Induction motors are the main workhorses of industry. Condition monitoring (CM) of motor based systems plays an important role in the early detection of possible defects, averting adverse operational and financial effects caused by unexpected breakdowns. Limited information has been found which explores the diagnostic abilities of voltage and motor current signals from motors with variable speed drives (VSDs), which are increasingly used in industry to obtain better dynamic response, higher efficiency and lower energy consumption. This study addresses the gap identified by carrying out a systematic review of the monitoring of mechanical systems using induction motors with sensorless VSDs. Specifically, existing techniques often prove ineffective for common internal and external faults that develop in Induction motors. The primary aim is to extract accurate diagnostic information from the power supply parameters of a VSD to monitor IM driven systems for early diagnosis of both mechanical and electrical faults. This thesis examines the effectiveness of both motor current and voltage signals using spectrum analysis for detecting broken rotor bar(s) and/or shaft misalignment and gear oil viscosity changes with different degrees of severities under sensorless control (close) modes. The results are obtained from common spectrum analysis applied to signals from a laboratory experimental setup operating under different speeds and loads. Evaluation of the results shows that the faults cause an increase in sideband amplitudes, which can be observed in both the current and voltage signals under the sensorless control mode. In addition, combined faults cause an additional increase in the sideband amplitudes and this increase can be observed in both the current and voltage signals. The voltage signals show greater change compared with the current signals because the VSD adapts the voltage supply source to compensate for changes in the system dynamics. Furthermore, this study also presents a model of an induction motorfed by a variable speed drive (VSD), as an approach to simulate broken rotor bars and shaft misalignments to give an in-depth understanding of fault signatures. The model was validated with experimental results in both current and voltage signals, with good agreement. The model confirmed that BRB causes a shift and increase in the amplitudes of the sidebands with the amplitudes of the rotor frequency components increased due to shaft misalignment.