Detection of Industrial Motor Fault using Signal Processing Algorithms

Induction motors are the most commonly used motor in petrochemicals, agrochemicals, pharmaceuticals, cement, steel, and many other process industries. A failure of any single component or sub-components of the induction motor can result in a plant shutdown. Hence, it is crucial to diagnose different types of faults in induction motors. The current study overviews a vibration-based condition monitoring method for detecting bent shaft failures in large industrial induction motors. This four-pole, 1.7-MW induction motor is primarily utilized in the petrochemical sector. A bent shaft is a typical problem with the motor's rotational portion. A bent shaft creates excessive vibration in a machine depending on the extent and position of the bend. This present study uses vibration data samples collected during the last four years. The data is collected from the deteriorating status of the induction motor. As the vibration signal of the corresponding data sets is non-stationary time-frequency domain analysis is mainly used. Several types of failures happened in various frequency sub-bands of an induction motor. As a result, the discrete wavelet transform is primarily utilized to detect defective frequency ranges. The objective of this study is to use discrete wavelet transform based vibration analysis to check if there is a trend in percentage energy and, if so, to identify the frequency range of the bent shaft issue of an induction motor. In discrete wavelet transform analysis, the detail sub-band-1 shows the deteriorating condition of an induction motor. The frequency range of detail sub-band-l is 1280 Hz to 2560 Hz. On the time-frequency plots, more precise detection of frequency range may be obtained. If the vibration is increased from the lowest to highest peak value in a time-frequency plot, the corresponding energy likewise attempts to locate compactly in discrete bits.