Date of Award

January 2017

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Mechanical Engineering

First Advisor

Cai Xia Yang

Abstract

This thesis focuses on failure analysis of rotating machines based on vibration analysis and signal processing techniques. The main objectives are: identifying machine’s condition, determining the faults specific response, creating methods to correct the faults, and investigating available statistical analysis methods for automatic fault detection and classification. In vibration analysis, the accelerometer data is analyzed in time and frequency domain which will determine the machine’s condition by identifying the characteristic frequencies of the faults. These fault frequencies are specific for each type of machine’s faults. Therefore, they are referred to as faults’ signatures. The most common faults of the rotating machines are unbalanced load torque, misaligned shaft, looseness, and bearing faults. The second objective is to find correction methods for rectifying the faulty situations. Therefore, correction methods for the unbalanced condition are comprehensively studied and a novel method for balancing an unbalanced rotor is developed which is based on image processing methods and results in lowering machine’s vibrations. Another objective of this research is to collect huge amount of vibration data and implement statistical data analysis methods to categorize different machine’s conditions. Therefore, principal components analysis, K-nearest neighbor, and singular value decomposition are implemented to identify different faults of the rotating machines automatically. The statistical methods have demonstrated high precision in classifying different faulty situations. Fault identification at early stages will enhance machine’s health and reduces the maintenance costs significantly. The statistical methods are easy to implement, and have disaffected the need for an expert maintenance engineer and will identify the machine’s fault automatically.

Download

Share

COinS