Abstract

Bearings are crucial components for the mechanical system that allows relative motion between two parts. Bearings primarily used to reduce the friction on the component. However, Bearings can fail for several reasons such corrosions, fatigue and electrical damage ..etc .This research aims to develop an accurate prediction model and visualization dashboard to avoid the costly downtimes, high repair cost and enhance the maintenance experience for workers and engineers. The data used in this research is CASE WESTERN RESRVE UNIVERSITY (CWRU) bearing data set. The data set contains signals vibration readings of bearings that can used for bearings fault detection. The method used for building the model is Cross-Industry Standard Process for Data Mining (CRISP-DM) methodology which is commonly used in machine learning, The algorithms used are SVM and RF. However, RF showed higher accuracy for fault classification. Then random forest for feature reduction used for knowing the main attributes, the algorithm showed that attributes sd, rms, mean, kurtosis and max are the top five attributes but that caused a minimal reduction in accuracy. The model using RF with all attributes achieved 95.7% while after the feature reduction achieved 95.6%. the second part form this research is to study the relationship between signals vibration with fault size and location. Moreover, fault located in the outer ring demonstrated higher values while the bearings working normally had the lowest signal vibration values.

Library of Congress Subject Headings

Bearings (Machinery)--Data processing; Machine learning; Data mining

Publication Date

11-2023

Document Type

Thesis

Student Type

Graduate

Degree Name

Professional Studies (MS)

Department, Program, or Center

Graduate Programs & Research

Advisor

Sanjay Modak

Advisor/Committee Member

Ehsan Warriach

Campus

RIT Dubai

Plan Codes

PROFST-MS

Download

Share

COinS