Genetic Algorithm based Neural Network for Classification of Rolling Element Bearing Condition

Vijay, G S and Pai, Srinivasa P and Sriram, N S (2012) Genetic Algorithm based Neural Network for Classification of Rolling Element Bearing Condition. In: National Symposium on Acoustics (NSA-2012), Dec 5 toDec 7 2012, Tamilnadu, India.

[img] PDF
paper_11_2012_1.pdf
Restricted to Registered users only

Download (4MB) | Request a copy

Abstract

In this paper, the performance of Multi Layer Perceptron Neural Network (MLPNN) based on features selected using Fisher's Criterion (FC) and Genetic Algorithm (GA) has been compared for rolling element bearing condition classification. Vibration signals have been acquired from four conditions (normal bearing, bearing with defect on inner race, bearing with defect on outer race and bearing with defect on ball element) of a deep groove ball bearing (6205) from a customized bearing test rig under different conditions of load and speed. Several statistical features in time domain and frequency domain were extracted from denoised vibration signal. GA was also used to optimize the learning rate, momentum factor and number of neurons in the hidden layer. MLPNN based on GA has been found to perform better than that based on features selected using FC, thereby helping in effective classification of rolling element bearing condition.

Item Type: Conference or Workshop Item (Paper)
Subjects: Engineering > MIT Manipal > Mechanical and Manufacturing
Depositing User: MIT Library
Date Deposited: 20 Mar 2013 11:33
Last Modified: 20 Mar 2013 11:33
URI: http://eprints.manipal.edu/id/eprint/79192

Actions (login required)

View Item View Item