Comparison of Dimensionality Reduction Techniques for Effective Fault Diagnosis of Rolling Element Bearing

Vijay, G S and Patil, Vijay M and Kumar, H S and Vishwash, B (2015) Comparison of Dimensionality Reduction Techniques for Effective Fault Diagnosis of Rolling Element Bearing. In: International Conference on Emerging Trends in Engineering (ICETE-15), 8th – 9th May, 2015, NMAM Institute of Technology, Nitte.

[img] PDF
http___rms.manipal.edu_PublicationEntry_DisplayPdf.aspx_val=__172.16.19.67_rmspdf_00000254_CP_21-07-16-04-15-29.pdf - Published Version
Restricted to Registered users only

Download (606kB) | Request a copy

Abstract

This paper uses Multi-Layer Perceptron Neural Network (MLPNN) for comparing the linear and non-linear dimensionality reduction techniques (DRTs) for fault diagnosis in rolling element bearing (REB).The vibration signals from normal bearing (N), bearing with defect on ball (B), bearing with defect on inner race (IR) and bearing with defect on outer race (OR) have been acquired under different radial loads and shaft speeds. These signals were subjected to Kurtosis-Hybrid Threshold rule (K-HTR) based denoising technique, from which 17 statistical features have been extracted. Linear DRTs namely, principal component analysis (PCA), Fisher’s discriminant analysis (FDA) and nonlinear DRT namely Kernel-fisher discriminant analysis (K-FDA) have been used to select the sensitive features. The selected features have been evaluated using MLPNN. Finally a comparison of Linear and non-linear DRTs based on MLPNN performance is presented.

Item Type: Conference or Workshop Item (Paper)
Subjects: Engineering > MIT Manipal > Mechanical and Manufacturing
Depositing User: MIT Library
Date Deposited: 22 Jul 2016 10:17
Last Modified: 22 Jul 2016 10:17
URI: http://eprints.manipal.edu/id/eprint/146687

Actions (login required)

View Item View Item