The most common machinethat we see around us today is the so called asynchronous machine, also knownas Induction Motor. It has become the most widespread motor in use today. Three phase squirrel-cage induction motors(IMs) are widely used in industry, particularly in high-tech domains due totheir high power-to-weight ratio, low price, and easy maintenance. However,there performance includes constraints such as electrical and mechanical faults.Consequently, a variety of condition-monitoring techniques have been developedto tackle these problems 1. There has been great amount of research in thefault diagnosis of electric motors to identify the fault, reduce maintenancecosts, and prevent the catastrophic failures.The major faults inelectric motor can be categorized in to eccentricity, stator fault, brokenrotor bar, bearing fault.
With recent increased application of electric machineinto harsh industrial and transportation applications, more precise conditionmonitoring of bearing fault in an electric machine is required for higherreliability and safety purposes. Among all electric motor faults, bearingfaults account for more than 40%-50% of all the motor failures. There areseveral techniques that can be employed to detect and diagnose the bearingfaults like Fast Fourier Transform (FFT), Short Time Fourier Transform (STFT),Wigner-Ville Distribution (WVD), the Envelope Analysis (EA), and Wavelet Transform(WT).
Fast Fourier Transform (FFT) is the most commonly used method for analysingstationary signals. Unfortunately, FFT-based methods are not suitable fornon-stationary signals. Short time Fourier transform (STFT) is a supplementarymethod to overcome the disadvantages of FFT. The problem with STFT is that itprovides constant resolution for all frequencies since it uses the same windowfor the entire signal.
Therefore, STFT is suitable for the quasi-stationarysignal analysis instead of real non-stationary signal analysis. Envelopeanalysis (EA) is a useful method for fault diagnosis inducing periodic shockssuch as gears and bearings and has been applied extensively for conditionmonitoring of machine components. However, a critical drawback of this techniqueis that it requires knowledge the resonance frequency and filtering band. Inorder to overcome these limitations, this work gives a new method for bearingfaults diagnosis based on Hilbert Transform (HT) and Discrete Wavelet Transform(DWT). The monitoring results indicate that the proposed method improves thebearing faults diagnosis compared to other common techniques.The paper is organisedfollow: in section II presents discussion about signal processing techniques.
In section III presents Feature extraction. In section IV discuss the k-nearestneighbor (KNN) algorithm for differentiate the motor condition. In section VResult are discuss. In section VI concludes our contributions are discussed,and we conclude with some remarks and future work. BlockdiagramFigure shows thecomplete block diagram of fault detection and diagnosis. In first stage voltageand current signal of 3 phase (A,B and C) measure using step down transformerand current probe.
Data acquition is important part of fault detection anddiagnosis. The acquired signal then passed to the signal processing block. Discretewavelet transform signal processing technique are used. Then it is passed tothe feature extraction.
Feature are extract using feature extraction. The signalis passed to the k nearest neighbour classifier. The KNN classifier are usedcondition of the motor.