ANN Based Fault Detection Scheme for Bearing Condition Monitoring in SRIMs using FFT, DWT and Band-pass Filters

Abstract

Heavy duty electrical drives employ slip ring induction motors (SRIMs) owing to their excellent starting and running performance characteristics. However, hazardous working environment prevalent in certain industries renders these SRIMs prone to a number of unwanted anomalies. Therefore, condition monitoring of the working behavior of SRIMs is indispensable for accomplishing substantial production with minimum downtime. In this regard, the present research work proposes an efficient and effective condition monitoring scheme for the detection of ball bearing damage, which is a frequent fault in a SRIM. This is done using stator current as a viable detection parameter. Fast Fourier Transform (FFT) and discrete wavelet transform (DWT) is employed for the design and realization of the proposed scheme along with state variable band-pass filters. This forms the three-tier approach for fault detection. Furthermore, artificial neural network (ANN) based pattern recognition is used as a fourth-tier confirmation of the presence or absence of bearing damages as a process of fault pin-pointing. Real-time validation of the aforementioned scheme is carried out in LabVIEW based laboratory interface. Hazardous working environment in certain industries renders the use of sophisticated equipment and machinery to be relatively unviable. Therefore, a simplistic approach is rather indispensible in the current scenario. Therein lays the novelty of the present research work. The entire design, analysis and further testing is carried out in MATLAB/ Simulink and LabVIEW based laboratory interface using a 5.5 kW, 3-phase, 415 V, 50 Hz SRIM.