Bearing fault diagnosis of an induction motor using artificial intelligence techniques

Bearings are important part in electrical machines. Bearing faults are more common type of fault occurring in electrical machines. Condition monitoring of bearings in electrical machines is an important task which aids in maintaining the bearing in a healthy state. The condition monitoring system ne...

Full description

Saved in:
Bibliographic Details
Main Author: Sampath, Priyanka
Other Authors: School of Electrical and Electronic Engineering
Format: Theses and Dissertations
Language:English
Published: 2019
Subjects:
Online Access:http://hdl.handle.net/10356/78559
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:Bearings are important part in electrical machines. Bearing faults are more common type of fault occurring in electrical machines. Condition monitoring of bearings in electrical machines is an important task which aids in maintaining the bearing in a healthy state. The condition monitoring system needs data to be acquired from the machine. Real time vibration and current signals in steady state as well as transient state have been acquired from the induction motor using a data acquisition system and stored in a computer. Various techniques of bearing fault analysis have been studied. The analysis of bearing faults has been done using frequency domain techniques and time-frequency domain techniques. Fast Fourier Transform (FFT) method is used for fault analysis in frequency domain. Time-frequency domain analysis techniques involves a combination of Discrete Wavelet Transform (DWT) and Empirical Mode Decomposition (EMD). DWT method is used to decompose the signal into small wave packets and EMD methods is used to obtain the Intrinsic Mode Function (IMFs) and the residues which are the high frequency and low frequency components. Random Vector Functional Links (RVFL) which is a type of neural network, has been used to obtain the probability matrix. Performance of different classifiers like Support Vector Machine, linear and quadratic discriminant, ensemble methods, decision trees have been compared. An accuracy of 99.5% has been achieved for classification of fault using vibration signal under steady state operation of the induction motor.