Condition Monitoring and Fault Diagnosis of Induction Machine Rotor Systems based on Instantaneous Electrical Signals Analysis

Induction machines (IMs) play vital roles in a wide range of industry applications. But the frequently happening faults--rotor system faults will reduce the performance and efficiency of IMs. The rotor system faults generally include broken rotor bar (BRB) fault and rotor eccentricity fault (REF). Condition monitoring (CM) and early fault detection based on instantaneous electrical signals (IES) are the essential to ensure the safety and performance of IMs. For IMs, both BRB faults and REF will induce the sideband components at their related fault frequencies in the current spectrum. The sidebands around supply frequency in the spectrum analysis are commonly used as the indicators for rotor system fault detection. For REF related sidebands are relatively far away to the supply frequency. However, for BRB, the fault related sidebands are near around the supply frequency together with the sideband amplitude increase as the load increases, which makes the BRB fault indicators are easily buried in the supply frequency leakage especially under light loads. Moreover, when IMSs are controlled by sensorless field orientated control (FOC) drives, it becomes more challenging to diagnose rotor system faults than by openloop drives because of the noisy power supply of FOC. Therefore, this research subject investigates the effective methods to monitor the condition of the rotor system. It firstly investigates the current models of healthy motor, motors with partial or fully breakage bars at different positions, and motor with different degree eccentricity and the abnormal load. It proposes the breakage position, the breakage level and the external load affect the amplitude of rotor system fault indicator. Furthermore, in view of the fact that the leakage of the frequency analysis methods can bury the rotor system fault indicators under light loads. This research also proposes two kinds of methods to accurately demodulate and extract the fault features for the CM and fault diagnosis of IM, which are 1) An energy operator, implemented from time domain and frequency domain respectively, and 2) A novel morphological gradient (NMG) of mathematical morphology (MM), implemented by changing signal shape. The performance and effectiveness of the proposed signal processing methods are first verified through simulation study and then verified by the experimental study with different fault severity under different loads. The research also proposes that not only the current components in the low frequency band, but also the current components in the higher frequency bands can be used for CM and fault diagnosis of rotor system. And this proposal is verified by experimental study