test Browse by Author Names Browse by Titles of Works Browse by Subjects of Works Browse by Issue Dates of Works

Advanced Search
& Collections
Issue Date   
Sign on to:   
Receive email
My Account
authorized users
Edit Profile   
About T-Space   

T-Space at The University of Toronto Libraries >
School of Graduate Studies - Theses >
Doctoral >

Please use this identifier to cite or link to this item: http://hdl.handle.net/1807/29916

Title: Early Fault Detection Scheme and Optimized CBM Strategy for Gear Transmission System Operating under Varying Loads
Authors: Yang, Ming
Advisor: Makis, Viliam
Department: Mechanical and Industrial Engineering
Issue Date: 31-Aug-2011
Abstract: The development of fault detection schemes and optimized condition-based maintenance (CBM) strategies for gear transmission systems has received considerable attention in recent years. Most models considered the gear systems operating under constant load. Constant load assumptions imply that changes in condition monitoring data are caused only by deterioration of the gear systems. However, most real gear systems operate under varying loads and speeds which affect the condition monitoring data signature of the system. This typically makes it difficult to recognize the occurrence of an impending fault and to optimize a CBM strategy. This thesis first presents a novel approach to detect and localize the gear failure occurrence for a gear system operating under varying load conditions. An autoregressive model with exogenous variables is fitted to the time-synchronously averaged (TSA) condition monitoring data when the gear transmission system operated under various load conditions in good state. The fault detection and localization indicator is calculated by applying F-test to the residual signals of the ARX model. Then, the gear deteriorating process is modeled as a three state hidden continuous Markov model with partial information, and the model parameters are estimated using ARX model residuals. The pseudo likelihood function is maximized and the EM algorithm is applied. Finally, a multivariate Bayesian process control is developed to optimize the decision variables of the CBM strategy: the time interval for next data collection and the control limit for the posterior probability calculated at next data collection epoch. The decision variables are calculated by minimizing the total expected cost from current stage to the end of the production run. Dynamic programming is applied to update the decision variables after new monitoring data are acquired, and the maintenance decision is made by taking all available information into account.
URI: http://hdl.handle.net/1807/29916
Appears in Collections:Doctoral

Files in This Item:

File Description SizeFormat
Yang_Ming_201106_PhD_Thesis.pdf2.7 MBAdobe PDF

Items in T-Space are protected by copyright, with all rights reserved, unless otherwise indicated.