Enhanced algorithm for motor rotor broken bar detection

Abstract

Motor rotor broken bar is one of the predominant failure modes of squirrel cage induction motors. There are numerous researched methods for identifying rotor bar faults: motor current signature analysis, acoustic noise measurements, vibration monitoring, temperature monitoring, electromagnetic field monitoring, infrared recognition, radio frequency emissions monitoring, etc. The most frequently used method is called the Motor Current Signature Analysis (MCSA). It is based on a signal analysis of the motor current, obtained via a regular current transformer used for motor protection purposes. It is difficult to detect rotor bar failures by looking into the currents waveform-time domain analysis, however impact of rotor broken bars to the stator currents can be determined by analyzing spectrum of frequency distribution in the frequency domain.Many factors affect reliable detection of the motor broken bar; motor load, system frequency and motor speed, construction of the motor etc. A new algorithm takes into account all these factors to adapt to a changing operational condition of the motor. Also, by learning the healthy motor frequency spectrum signature, the detection of a broken rotor bar can be made even more deterministic.The new algorithm was extensively tested on the induction motors with different system and motor conditions-results of this testing are presented. Lessons learned from the field installations are presented as well.I.

Advancement of Fault Diagnosis and Detection Process in Industrial Machine Environment

“…Eccentricity related faults in rotor cause a consistent air gap that corresponding the unbalanced magnetic force and non-consistent air gap magnetic field which producing the vibration and noise on stator [47][48][49]. Most of the research [50][51][52][53][54] has been published in last years about the vibration and noise on Permanent magnet synchronous (PMS) and DC motors but little research has been done on the eccentricity related fault in rotor to observe the vibroacoustic behavior of IM motors.…”

“…These parameter changes depend upon the severity of the fault and interaction with other parameters. A broad range of survey [34][35][36][37][38][39][40][41][42][43][44] has been done from different researchers and many techniques (e. g. thermal monitoring [50,51], current analysis [47,48], torque monitoring [54,55], noise monitoring [52,53] and vibration analysis [49]) for fault detections has been investigated, but all other techniques except MCSA are required some expensive sensors for monitoring machine condition [20,45]. This is the reason the current monitoring techniques is non-intrusive and also may use for the remote monitoring in the central motor control center.…”

“…This is the reason the current monitoring techniques is non-intrusive and also may use for the remote monitoring in the central motor control center. Several IM Faults diagnoses techniques are based on the Fast Fourier Transform (FFT) spectrum signature analysis using the stator current [47]. The methods used to detect the BRB fault in IM can be classified into two categories, invasive and non-invasive method [56].…”

“…Basically, this method is based on the produced sidebands around the fundamental supply frequency in stator current [57]. Several works [58][59][60] has been reported in previously published papers for the development of machine condition monitoring application using MCSA among different industrial case studies [47]. The MCSA use the current spec-trum of machine for finding the required attributes of fault frequencies at what time a fault present.…”

“…Hence, there is a need to identify and select the most valuable features which are very important in machine fault identification and localization process. To use the complete set of feature sets will cause much cost of time and low performance because in the original features set there are numerous irrelevant and laid off features existed [47].…”