Date of Award

2001

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Industrial Engineering

First Advisor

Gerald M. Knapp

Abstract

This research presents the creation of a new model for automating the generation of component and system reliability estimates from simulated field data for tightly coupled systems. The model utilizes the CMAC neural network architecture, which resembles the human cerebellum and is capable of approximating nonlinear functions. An analysis and testing of the network as a tool for reliability prediction of dependent components within an assembly has been performed. In order to evaluate the performance of the model, the network has been tested on simulated data and provided over 90% performance accuracy in learning non-linear functions that represent the dependency between components. This serves as a valuable tool for maintenance personnel faced with important and costly decisions regarding equipment maintenance policies.

ISBN

9780493328560

Pages

106

Share

COinS