AbstractAs an essential part of motor-driven systems, induction motors (IM) play a vital role in a wide range of industry applications. This is mainly because of their simple design in the rotor and stator which serves as a powerful, rugged, and extremely durable structure, with excellent durability. However, these motor-driven systems are susceptible to premature faults under harsh and complex working conditions. These include both electrical and mechanical faults such as broken rotor bars, stator resistance imbalance, gear tooth breakage, gearbox oil shortage, shaft misalignment as well as combined faults. These faults can cause system damage, influence production activities, and consume additional energy which eventually have an impact on production and cause economic loss. As a result of precise diagnosis and early identification of incipient defects, unscheduled maintenance is performed quickly.
Condition monitoring has proven to be important for detecting faults early, to avoid any operational collateral damage and reduce the financial obligation incurred through these faults. A proper diagnostic approach should collect the barest minimum measurement from a motordriven system and extract a diagnosis through signal analysis, allowing its state to be inferred while providing a clear indication of impending failure modes in the shortest amount of time.
Proper condition monitoring can be done using current, vibration, acoustic emission, instantaneous angular speed, thermal etc. However, some of these techniques have proven
ineffective for detection and diagnosis. Meanwhile, increased noise levels and constant variations in power supply characteristics often impede the detection of a fault. Therefore, this study focused on the use of motor current to detect and diagnose faults because of its costeffective, reliable, accurate, and non-intrusive monitoring capability at a remote location from the equipment in hostile environments and its easy implementation.
Numerous methods have been developed to analyse any conceivable incipient fault. A common analytical technique such as motor current signal analysis (MCSA) has been used in recent years and has proved to be difficult in detecting deliberate signals that prevail with noise around the supply frequency because of its small modulations around the fault frequency. Hence, this study evaluated four different techniques on several issues in the realisation of motor-driven system diagnostics, using motor current signals under variable load and speed operations. Thereafter, MSB became identified as the signal-processing technique with the greatest potential to meet the assessment due to its unique properties of simultaneous noise reduction and modulation enhancement.
To achieve the development of a more efficient and reliable diagnostic technique for motordriven systems using the motor current signals, a numerical model was proposed, which includes IM and gearbox model. This model can be used to simulate the faults happening in a motor gear system and to predict motor current signals based on the fault signature features. A series of simulation studies and experimental validation studies were carried out to investigate the causes of motor-driven system faults on the characteristics of electrical signals such as current, voltage and power. Then MSB analysis was applied to extract the fault characteristics features.
In addition, performance analysis of MSB showed that motor currents and voltage signals can be useful. Results demonstrate that BRB, SRI, and compound faults increase the sidebands in electrical signals, which can be detected. Furthermore, MSB diagnosis is very effective in detecting different oil shortages and gear tooth breakage. Results show that the characteristic shaft frequencies 𝑓! ± 𝑓"# are good diagnostic information.
In conclusion, investigation of the motor-gear dynamic model helped in the prediction of motor current signals for fault diagnosis. The test outcomes show that MSB has a superior execution in separating spectrum amplitudes due to fault severities and thus delivers better diagnosis performance.
|Date of Award
|27 Feb 2023
|Fengshou Gu (Main Supervisor) & Andrew Ball (Co-Supervisor)