Author: Marta Manzano

Title: DYNAMICS AND HAZARDS OF CASCADING FAILURE IN CELL ARRAYS: ANALYSIS, PASSIVE MITIGATION, AND ACTIVE SUPPRESSION

Defense date and time: Friday- Feb 28, 2020 at 10 am

Location: A. James Clark Bldg.  (AJC 2132)

Committee members:

Title: Deep Adversarial Approaches in Reliability 

Date: Tuesday, December 17th

Time: 1pm

Location: AV Williams 2460 ECE Conference Room 

Committee Members:

Professor Mohammad Modarres, Chair

Associate Professor Enrique Lopez Droguett

Assistant Professor Mark Fuge

Assistant Professor Katrina Groth

Professor Balakumar Balachandran

Professor Mohamad Al-Sheikhly (Dean’s Representative)

Abstract:

Reliability engineering has long been proposed with the problem of predicting failures using all the available data.  As modeling techniques have become more sophisticated, so too have the data sources from which reliability engineers can draw conclusions.  The Internet of Things (IoT) and cheap sensing technologies have ushered in a new expansive set of multi-dimensional big machinery data in which previous reliability engineering modeling techniques remain ill-equipped to handle.  Therefore, the objective of this dissertation is to develop and advance reliability engineering research by proposing four comprehensive deep learning methodologies to handle these big machinery data sets. In this dissertation, a supervised fault diagnostic deep learning approach with applications to the rolling element bearings incorporating a deep convolutional neural network on time-frequency images was developed. A semi-supervised generative adversarial networks-based approach to fault diagnostics using the same time-frequency images was proposed.  The time-frequency images were used again in the development of an unsupervised generative adversarial network-based methodology for fault diagnostics.  Finally, to advance the studies of remaining useful life prediction, a mathematical formulation and subsequent methodology to combine variational autoencoders and generative adversarial networks within a state-space modeling framework to achieve both unsupervised and semi-supervised remaining useful life estimation was proposed.

All four proposed contributions showed state of the art results for both fault diagnostics and remaining useful life estimation. While this research utilized publicly available rolling element bearings and turbofan engine data sets, this research is intended to be a comprehensive approach such that it can be applied to a data set of the engineer’s chosen field. This research highlights the potential for deep learning-based approaches within reliability engineering problems.

Title: EIT BASED PIEZORESISTIVE TACTILE SENSORS: A SIMULATION STUDY

ABSTRACT:

Title: Modeling Syndromic Surveillance and Outbreaks in Subpopulations

Date: Monday, Dec. 16, 2019

Time: 12:00pm

Location: Martin Hall  EGR-2164

Committee Members:

Professor Jeffrey Herrmann

Professor Robert Gold
Assistant Professor Allison Reilly

Professor Linda Schmidt

Assistant Professor Monifa Vaughn-Cooke

Abstract:

This research is motivated by the need to assist resource limited communities by enhancing the use of syndromic surveillance (SyS) systems and data. Public health agencies and academic researchers have developed and implemented SyS systems as a pattern recognition tool to detect a potential disease outbreak using pre-diagnostic data. SyS systems collect data from multiple types of sources: absenteeism records, over the counter medicine sales, chief complaints, web queries, and more. It could be expensive, however, to gather data from every available source; subsequently, gathering information about only some subpopulations may be a desirable option. This raises questions about the differences between subpopulation behavior and which subpopulations’ data would give the earliest, most accurate warning of a disease outbreak.

To investigate the feasibility of using subpopulation data, this research will gather and organize SyS data by subpopulation (separated by population characteristics such as age or location) and identify how well the SyS data correlates to the real world disease progression. This research will study SyS how reports of Influenza-like-illness (ILI) in subpopulations represent the disease behavior. The first step of the research process is to understand how SyS is used in environments with varying levels of resources and what gaps are present in SyS modeling techniques. Various modeling techniques and applications are assessed, specifically the Susceptible Infected Recovered “SIR” model and associated modifications of that model. Through data analysis, well correlated subpopulations will be identified and compared to actual disease behavior and SyS data sets.  A model referred to as ModSySIR will be presented that uses real world community data ideal for ease of use and implementation in a resource limited community. The highest level research objective is to provide a potential data analysis method and modeling approach to inform decision making for health departments using SyS systems that rely on fewer resources.