Misalignment-Related Defect Detection using Discrete Wavelet Transform

Abstract

Induction motors are most commonly used in many industries, including petrochemicals, oil, and steel. A single failure in any of the induction motor's components or sub-components can result in a plant shutdown. The plant will suffer significant financial losses as a result. It is crucial to diagnose different types of faults in induction motors. Various condition monitoring techniques diagnose faults in induction motors in the early stages. Vibration analysis is most commonly used among different condition monitoring techniques due to its higher accuracy than other methods. Vibration analysis is used to detect various types of faults in induction motors. The acceleration vibration data corresponding to multiple types of defects are gathered from publicly available web resources. The primary objective of this research work is to explore the severity of horizontal and vertical misalignment defects utilizing a signal processing approach. To achieve this objective, Discrete Wavelet Transform (DWT) is used to detect abnormal behavior of the induction motor. The Daubechies-4(db4) wavelet is chosen as a mother wavelet. As Daubechies wavelet is an orthogonal wavelet, the percentage energy in all decomposed sub-band will equal the original energy of the signal. The energy level of sub-bands is compared with the healthy condition of the motor to detect significant changes in motor fault.