Date of Award

Spring 2015

Document Type


First Advisor

Esterline, Albert


This thesis reports on work in machine learning and high-performance computing for structural heath monitoring. The data used are acoustic emission signals, and we classify these signals according to source mechanisms, those associated with crack growth being particularly significant. The work reported here is part of a larger project to develop an agent-based structural health monitoring system. The agents are proxies for communication- and computation-intensive techniques and respond to the situation at hand by determining an appropriate constellation of techniques. The techniques thus structured are executed by a workflow engine, which is part of the contribution reported here. It is critical that the system have a repertoire of classifiers with different characteristics so that a combination appropriate for the situation at hand can generally be found. The classifiers are trained using machine-learning techniques, and we report on investigations we conducted on three supervised and two unsupervised learning techniques to determine which techniques are the best to use in a particular situation.