Detection of Broken Rotor Bar Fault in Induction Motor Using Paramter Estimation Approach

The detection of faults in a dynamic system can help to avoid severe damage where system reliability and safety are essential. An induction motor can fail either due to internal faults such as short circuit of motor leads; broken rotor bar and cracked rotor end rings or by external faults such as phase failure, asymmetry of main supply and mechanical overload (Karami et al 2010). Detection and monitoring of rotor bar breakage in squirrel cage induction motors has been a well studied problem. The widely studied method is the motor current signature analysis (MCSA) in which certain side band frequencies of the current spectrum are used to identify broken rotor bar (Jung et al 2006, Kumar et al 2009). This method generally suffers from load disturbance and speed variation (Kim and Seok 2010). Several new techniques are also reported in the literature to identify and analyse stator current signature such as neural network (Arabaci and Bigin 2009, Moradian et al 2006) and support vector machine classifiers (Armaki and Roshanfekr 2010). Thomas et al (2003) has shown that the characteristic frequencies of a particular fault can be obtained from the air-gap torque spectrum. Other techniques such as instantaneous angular speed (IAS) based rotor bar fault detection are also reported (Ahmed et al 2006).