Engineering Commons^™

Improving The Robustness Of Compressed Deep Learning Models Against Class Imbalance, Baraa Saeed Ali

Wayne State University Dissertations

Deep Learning (DL) models are deployed ubiquitously, as they power a wide range of critical applications, including image classification, fraud detection, autonomous vehicles, robots, and NLP. However, their massive size and huge memory footprint (overparameterization) represent a serious challenge to the efficient deployment of such models, especially in resource-scarce environments such as wearable devices, smartphones, edge devices, and embedded systems. Therefore, model compression techniques are typically used to shrink the model size to the currently available computational and memory budget and to accelerate training and inference without sacrificing model accuracy and performance. Therefore, the DL research community considers model compression …

A Strategic Infrastructure Improvement Framework For Intermodal Transportation Networks, Ayoub Abusalih

Graduate Theses, Dissertations, and Problem Reports (ETD)

In this research, we propose a novel approach to design infrastructure networks for intermodal freight transportation systems, which incorporates railways, highways, and inland waterways (IWW). The objective of our study is to identify the optimal set of hubs to be built and operated over an extended time, based on the projected domestic cargo demand. Unlike traditional hub location models, our approach introduces hybrid hubs, where hybrid transportation modes are integrated to facilitate cargo handling. This innovative integration enables more efficient intermodal connections, leading to tangible reductions in operating costs, and carbon emissions. Specifically, we propose a mixed integer programming model …

Hydrogen-Enriched Lng As A Mid-Term Solution To Mitigate Greenhouse Gas Emissions From Shipping, Kang-Ki Lee

World Maritime University Ph.D. Dissertations

In light of the International Maritime Organization’s (IMO) ambitious Initial GHG Strategy, the rise in international shipping’s carbon dioxide emissions by 2 percent in 2022 compared to 2019 poses a formidable challenge. This increase underscores the pressing need to address the limited availability of green hydrogen, prompting the exploration of mid-term solutions to bridge this critical gap. The IMO’s 2023 Strategy on Reduction of GHG Emissions from Ships attempted to tackle this challenge head-on, recognizing the urgency of reducing the industry’s substantial carbon footprint. While hydrogen offers a simple solution as a zero-carbon fuel, LNG remains the most widespread alternative …

Modeling Neighborhoods As Fuel For Wildfire, Bryce Alan Young

Graduate Student Theses, Dissertations, & Professional Papers

Wildfire models drive billions of dollars in risk mitigation efforts. However, the modeling community currently lacks a representative fuelscape on which to base simulations of fire spread in the built environment and the wildland-urban interface (WUI) where vegetation and structures act together as fuel for wildfire. This thesis advances wildfire risk modeling by addressing the underdeveloped representation of the built environment in existing frameworks. By identifying inconsistencies in how structure and defensible space features are defined and used across empirical studies, predictive indices, and fire spread models, this research lays the groundwork for standardized modeling approaches and feature selection (Chapter …

Multi-Source Remote Sensing–Based Soil Moisture Prediction Using Machine Learning, Niraj Neupane

Electronic Theses and Dissertations

Soil moisture (SM) plays a central role in climatic and environmental processes, influencing shear strength of soil, agricultural productivity, land–atmosphere interactions, and hydrologic functioning. However, accurately estimating SM across diverse climatic regions remains challenging due to spatial heterogeneity, limited in situ measurements, and inconsistencies in sensor resolution. Machine learning (ML) and remote sensing offer promising avenues for improving SM prediction, yet many existing approaches struggle with generalization across climatic gradients and often fail to capture temporal variability. This study integrates multi-source satellite and climate datasets, including SMAP L4_SM, MODIS land surface temperature, Daymet meteorological variables, and in situ observations from …

Quantitative Assessment And Validation Of Thermal Profiles In The Protodune-Ii Horizontal Drift Detector Using Computational Fluid Dynamics, Hunter Wallster

Electronic Theses and Dissertations

The Deep Underground Neutrino Experiment (DUNE) aims to advance understanding of neutrino properties using large liquid Argon time projection chambers (LArTPCs). The ProtoDUNE-II Horizontal Drift (HD) detector, constructed at CERN in Switzerland, serves as 1:20 scale prototype to validate design and operational parameters for DUNE’s Far Detectors. Accurate prediction of cryostat thermal behavior is critical to predicting liquid Argon purity distributions and detector performance, motivating the development of a validated computational fluid dynamics (CFD) model. This research extends previous work at South Dakota State University by constructing a higher-fidelity CFD simulation of ProtoDUNE-II HD using Siemens Simcenter StarCCM+® 19.04.007. Geometry …

Mid-Scale Rover Gravity Offloader (Mrgo): Design, Implementation, And Experimental Validation, Alexander Schaar

Electronic Theses and Dissertations

This thesis presents the design, implementation, and validation of the Mid Scale Rover Gravity Offloader (MRGO), a terrestrial testbed developed to simulate partial-gravity conditions for planetary rover testing within a three-dimensional environment. The MRGO enables consistent, user-defined offloading force as a rover drives freely within a 20 by 18 ft test area, while a stepper motor rail gantry autonomously tracks the rover’s motion and maintains overhead alignment. Mounted to the moving gantry is a vertical carriage column that houses the offloading mechanism, which accommodates up to 5 ft of vertical displacement without loss of force accuracy. This configuration allows realistic …

Lorawan-Enabled Iot Solution For Smart Farming, Talha Khan

Electronic Theses and Dissertations

The rising global demand for food, driven by population growth, alongside a declining rural workforce, presents a great challenge for agriculture. Precision agriculture, which relies on the widespread adoption of smart farming technologies, is anticipated to enhance agricultural productivity and sustainability. Emerging technologies, such as machine vision, artificial intelligence (AI), and the Internet of Things (IoT), offer innovative and promising solutions to address these challenges. This study focuses on LoRaWAN, a prevalent Low-Power Wide-Area-Network (LPWAN) IoT technology, examining its application in crop and livestock farming as well as its associated cybersecurity challenges. LoRaWAN IoT systems were constructed for three application …

Efficient And Test-Time Adaptive Visual Object Tracking In The Wild, Ram J. Zaveri

Graduate Theses, Dissertations, and Problem Reports (ETD)

Tracking a single object, given the location at the first frame, has been an ongoing challenge in the vision community for decades. Most recent approaches provide reasonably good performance, especially when benchmarked on in-distribution (ID) datasets, i.e., on the testing portion of the same datasets used for training. However, they incur high computational costs and hardware constraints, making their deployment in the wild for mobile, autonomous, and IoT applications still challenging. Efficient visual trackers address the efficiency aspect of such bottlenecks; however, they tend to overfit to their training distributions and lack generalization abilities, resulting in them performing well on …

Influence Of Actuated Air Brakes On The Apogee Of Sub-Orbital Rockets, Michael L. Farha

Graduate Theses, Dissertations, and Problem Reports (ETD)

Collegiate engineering competitions have played a foundational role in the educational outcomes of university students for as long as they have existed. Among these competitions is the Spaceport America Cup, which challenges students to design, build, and launch high-power rockets with the incentive to fly as close as possible to a pre-established target apogee. Many teams competing in this competition have chosen to approach the challenge by incorporating deployable air brakes into their rocket designs. This approach comes with many uncertainties, both in controls and aerodynamics. For teams just starting out, aerodynamic performance is often one of the most challenging …

Physics-Informed Neural Network Based Aerodynamic Modeling Framework, Nathaniel E. Michek

Graduate Theses, Dissertations, and Problem Reports (ETD)

Significant advances have been made in developing aerodynamic models over the years. While these advances span many modeling techniques and data collection methods, certain aerodynamic regimes still pose significant challenges. These regimes commonly occur when an aerodynamic body is under extreme flight conditions. Many of these conditions occur simultaneously in the rare and poorly understood case of a tumbling aerodynamic body. The flight of a tumbling aerodynamic body goes through the entire range of aerodynamic angles, alpha and beta, causing significant non-linearities and time-dependent effects associated with flow separation. During tumbling, the aerodynamic body simultaneously rotates about all three axes …

Influence Of Deck Area And Average Daily Traffic On The Deterioration Of Bridge Superstructures., Faysal Ahamed

Graduate Theses, Dissertations, and Problem Reports (ETD)

State Departments of Transportation (DOTs) are responsible for maintaining approximately 623,000 bridges across the United States. Effective bridge management requires a comprehensive understanding of the factors influencing structural deterioration. Numerous studies have explored the effects of average daily traffic (ADT) and deck area (DA) on bridge deck deterioration. Given the interdependence between bridge decks and superstructures, it is plausible that ADT and DA significantly influence superstructure condition ratings and deterioration trends. However, their impact on superstructures remains insufficiently studied. In addition, existing studies primarily focus on overall bridge deterioration without adequately differentiating between superstructure designs and materials and maintenance authorities. …

Biomechanics Of Teeth Alignment Using Clear Aligners With Various Attachment Shapes And Orientations, Egon Mamboleo

Graduate Theses, Dissertations, and Problem Reports (ETD)

Clear aligners have emerged as the most popular and preferred method of treatment for patients with orthodontic malocclusions. This is greatly due to the comfort and aesthetically appealing factors when compared to fixed appliances . Clear aligners are either thermoformed or 3D direct-printed plastics that apply biomechanical forces to the surface of teeth to trigger tooth movement and bone remodeling process. Common Class I malocclusions with mild or moderate crowding can be treated with clear aligners alone. However, treatment of Class II and Class III malocclusions that require extraction of permanent teeth or correction of severe rotations, and teeth extrusions …

Development And Evaluation Of Machine Learning Models For Early Pediatric Sepsis Prediction, Ancita M. Andrade

Browse all Theses and Dissertations

Sepsis is a leading cause of pediatric mortality, claiming more lives in the United States annually than all childhood cancers combined. Early identification in Emergency Departments (EDs) remains challenging, as the current Phoenix criteria establishes an updated international consensus definition for sepsis, however is not designed for use as a screening tool. This study aimed to develop predictive models identifying pediatric patients at risk of sepsis within 24 hours of admission. Multiple tree-based and deep learning models were trained utilizing clinical and laboratory data from the initial four hours of presentation. Both the LightGBM and LSTM architectures demonstrated superior performance, …

Cellular Mechanisms Of Spinal Motoneuron Hypoexcitability Underlying Dynapenia In Aging, Ibrahim Abdul Halim

Browse all Theses and Dissertations

Age-related weakness remains poorly understood as the underlying mechanisms remain unclear. While synaptic input and muscular changes have been investigated with age, intrinsic motoneuron excitability alterations are often overlooked. This thesis provides the first direct assessment of intrinsic excitability and ion channel properties of spinal α-MNs from male and female mice across three ages: young, middle aged, and old. Our findings reveal a decline in intrinsic excitability of motoneurons with age in both sexes. Mechanistic analysis shows sex specific differences: female motoneurons exhibit increased dendritic size, hyperpolarized RMP, and SK channel overactivation, whereas males show only SK overactivation with age. …

Wearable Sensor Data Analysis For Machine Learning-Based Detection Of Posture And Autonomic Responses, Chaitanya Vardhini Anumula

Browse all Theses and Dissertations

This study investigates how Iyengar yoga postures influence autonomic nervous system (ANS) activity by analyzing multimodal physiological signals collected via wearable sensors. The physiological mechanisms underlying Iyengar yoga’s therapeutic effects remain under-explored at the granular, pose-level. Using data collected from 16 participants, this research evaluates whether machine learning models can distinguish between baseline, parasympathetic-dominant, and sympathetic-dominant states based on wrist-worn sensor data. The goals were to explore whether subtle postural variations elicit measurable autonomic responses and to identify which sensor features most effectively capture these changes. Participants performed a sequence of yoga poses while wearing synchronized sensors measuring electrodermal activity …

Impact Of Graph Structures For Rag Outcomes In Llms, Chris Davis Jaldi

Browse all Theses and Dissertations

Explainability, interpretability and adaptability (EIA) remain three central motivations for next-generation Artificial Intelligence (AI), especially as Large Language Models (LLMs) continue to engage with ever-increasing knowledge bodies. As the landscape pushes toward controllable agentic Retrieval-Augmented Generation (RAG) systems where AI agents engage in iterative, guided reasoning, a critical question arises as to the extent to which the knowledge design itself shapes these models' reasoning behavior. This work conducts a systematic evaluation of how different conceptualizations and representation of the identical knowledge affect an LLM's path-based reasoning capabilities. Through the introduction of controlled variations along graph structural complexity, linguistic and semantic …

Re-Parameterizing Adversarial Reprogramming In Low-Dimensional Subspace For Efficient Software Vulnerability Detection, Hootan Alavizadeh

Browse all Theses and Dissertations

Software vulnerabilities are a major cause of security breaches, making effective detection critical. Traditional learning-based methods require large datasets and significant computational resources, which are often impractical due to high annotation costs and data scarcity. To address this, we propose an innovative system, RearVul, which Re-parameterizes adversarial reprogramming in a low-dimensional subspace for software vulnerability detection. Unlike conventional approaches, RearVul repurposes a pre-trained classification model using adversarial reprogramming, enabling detection with minimal modifications. It learns a universal perturbation applied to program representations, preserving the original model’s feature extraction capabilities while adapting it to a new domain. Furthermore, we introduce a …

Subjective Readiness Forecasting Using Supervised Machine Learning And Wearable Device Data, Nathaniel Michael Weiland

Browse all Theses and Dissertations

Recent advances in wearable technology allow continuous monitoring of physiological and behavioral data, opening new opportunities for real-time assessments of readiness and well-being. However, creating predictive models that generalize across diverse users remains challenging, especially in high-stakes settings like the military, where preventable injuries, illnesses, and stress-related performance declines are frequent. This research assesses the feasibility of using supervised machine learning models trained on wearable device data to predict subjective readiness indicators—recovery, stress, injury, and illness. Data from over 10,000 users in the OHWS (Optimizing the Human Weapons System) program combined daily check ins with physiological metrics from Garmin, Polar, …

Computational Assessment Of Vitrimers Self-Healing For Renewable Energy And Aerospace Structures, Walaaeldin Mohamed Ahmed Derbala

Browse all Theses and Dissertations

Self-healing polymers, particularly vitrimers, are emerging as promising candidates in the development of advanced materials for renewable energy and aerospace structures. These materials exhibit dynamic covalent bond exchange mechanisms that enable reprocess ability, damage repair, and extended operational lifetime under harsh conditions. This study presents a density functional theory (DFT)-based computational investigation of the mechanistic pathways and energetics of bond exchange reactions in model vitrimer systems. We explore transition states, energy barriers, and thermodynamic features corresponding to associative and dissociative self-healing reactions in vitrimers. The study focuses on Diaminodiphenyl disulfide (AFD), a bifunctional molecule composed of two para-substituted aminophenyl rings …

Ab Initio Simulations For Oxidation Of An Ultra-High Temperature Ceramic (Hfb2), Wesley I. Black

Browse all Theses and Dissertations

Ultra-High Temperature Ceramics are a class of ceramics that possess high strength and melting points in excess of 3000°C. These ceramics are promising for aerospace applications, where materials need to endure high-temperature and high-stress environments. Utilizing ab initio simulations, this thesis research focuses on the atomistic details of oxidation for a typical ultra-high temperature ceramic material, namely hafnium diboride. The simulations provide energy barriers for transition from the initial to final structures via transition states on the (0 0 0 1) surface. These result in estimates for reaction rates and other thermodynamic features that are essential for assessing applicability under …

Reinforcement Learning For Adversarial Environments: Multi-Agent Hide And Seek With Multi-Modal Sensing, Christian Alejandro Carrizales

Browse all Theses and Dissertations

The development of intelligent and competitive agents in AI versus AI adversarial environments was explored through the utilization of reinforcement learning techniques with sensing modalities. A Hide-and-Seek simulation environment was developed using the Unity game engine along with the ML-Agents Toolkit. An engagement test campaign with a set of performance metrics was designed. Four AI versus AI adversarial scenarios were considered using the Proximal Policy Optimization (PPO) and Soft Actor-Critic (SAC) multi-agent reinforcement learning algorithms. Furthermore, the impact of sensing modalities on competing agents’ learning performance was investigated by varying the sensing capabilities of the hider and seeker, respectively. Experiments …

Dataset Generation For Routing Policy Study In Ad Hoc Wireless Networks, Vishnu Vishnu Priya

Browse all Theses and Dissertations

Ad Hoc wireless networks, with their decentralized architecture and dynamic topology, present challenges in reliable and energy-efficient routing. While machine learning (ML) and reinforcement learning (RL) offer promising solutions, progress is limited by the lack of realistic, high-fidelity datasets. This research introduces a simulation-based framework for generating four diverse datasets representing combinations of node mobility (mobile vs. static) and spatial distribution (random vs. clustered). Each dataset captures critical metrics such as Signal-to-Interference-plus-Noise Ratio (SINR), bottleneck rate, and power consumption across multi-hop paths. A lookahead-based greedy routing algorithm with scenario-aware power control is implemented to emulate practical behavior. Supervised ML models, …

Isometric Centroid Encoder (Ice) And Synthetic Data Generation Approaches For Biological Datasets, Prathyusha Kanakamalla

Browse all Theses and Dissertations

This thesis addressed two main challenges in biological data analysis: structure-preserving dimensionality reduction and synthetic data generation for small sample datasets. I proposed the Isometric Centroid Encoder (ICE), a supervised dimensionality reduction method that preserves pairwise distances between class centroids during dimension reduction. Unlike existing methods like Centroid Encoder and Super Encoder, ICE explicitly maintains geometric relationships between biological classes, achieving nearly perfect structure preservation at C dimensions (where C equals the number of classes) with strong performance even in 2D and 3D spaces. Additionally, I compared three generative models (VAE, LSH-GAN, and scDiffusion) for synthetic data generation on small …

Hydrogen Storage Density And Adsorption Energy Barriers On Li-Decorated Bc3 Nanosheet, Sri Venkat Pavan Upasi

Browse all Theses and Dissertations

Hydrogen is considered an emerging carrier of clean energy with renewable capabilities. Widespread hydrogen energy utilization necessitates efficient storage strategies. Functionalized nanomaterials, such as Li-decorated BC3 nanosheets, are among the primary candidate materials for hydrogen storage. This research investigates the feasibility of hydrogen storage by estimating energy barriers and their dependence on storage density on Li-decorated BC3 nanosheet. Density functional theory (DFT) simulations provide estimates of adsorption energies and saddle points for hydrogen storage and compare corresponding reaction rates. The results are expected to help us understand the advantages and possible shortcomings of hydrogen storage on such nanomaterials.

Mid-Wave Infrared Imaging Of Supersonic Combustor Exhaust Flow, Nathan Childs

Browse all Theses and Dissertations

A study was completed on the application of mid-wave infrared (MW-IR) imaging for diagnostics in supersonic combustion exhaust flows, with the objective of enhancing optical access and measurement accuracy. Other optical based techniques of thermography require complicated setups and analysis to determine the temperature of a flow with high accuracy, where MW-IR is a more simplistic "point and shoot" technique. The simplicity of MW-IR comes with the trade off of gaining simplicity but adding uncertainty into the measurements. The MW-IR camera was positioned to view the exhaust of the combustor to provide an unobstructed view of the flow, addressing limitations …

Laminar Separation Control Of An Eppler 387 Airfoil, Vincent R. Sheeler

Browse all Theses and Dissertations

A variety of aerodynamic devices operate at low Reynolds number conditions, such as unmanned aerial vehicles and low-pressure turbines in gas turbine engines. At low Reynolds numbers, many airfoils experience laminar boundary layer separation as the fow lacks the energy to overcome the adverse pressure gradient. Researchers have documented a variety of methods which can suppress laminar separation, and now focus on ways to reduce energy requirements to provide efective fow control. Aspects of fow control strategy such as actuator location and pulsing at frequencies which exploit natural instabilities in the fow can reduce energy requirements. In a study by …

Using The Historical Equity Action Lens (Heal) To Identify And Remedy Transportation Inequities From The Akron Innerbelt Project, Olivia Lane

Williams Honors College, Honors Research Projects

This project will utilize the Historical Equity and Action Lens (HEAL) to address transportation inequities in marginalized communities by incorporating historical and cultural insights into data collection and analysis. Specifically, the project will focus on the Akron Innerbelt, an infamous highway project that significantly impacted historically Black neighborhoods. By examining the historical, economic, and social context of the Innerbelt, the project will identify the project's long-term impacts Akron communities. The goal is to use this knowledge to inform improve transportation equity in the affected areas by developing a plan for the Innerbelt since its vacancy in 2016. The project will …

Mobile Weather Satellite Receiver, Luke Datsko, Sam Watts, Jason Do, Adam Bechtler

Williams Honors College, Honors Research Projects

The "Mobile Weather Satellite Receiver" project aims to create a portable, user-friendly device that receives and displays weather information from geostationary satellites, addressing the limitations of traditional weather sources like the Internet and weather radio, particularly in remote areas. This device will collect and demodulate satellite data, including imagery and Emergency Managers Weather Information Network (EMWIN) forecasts, to provide users with detailed local forecasts and real-time alerts.

Designed with a user-centric approach, the system includes a satellite dish, Software Defined Radio (SDR), a Raspberry Pi, and a custom software interface for ease of use. Its portability and ability to function …

An Improved Method To Protecting Skin Graft Dressings Following Surgery In Severe Burn Patients, Andrew Martin, Matt Flaker, Hailey Essinger

Williams Honors College, Honors Research Projects

Severe burns, including deep second- and third-degree burns, affect over 450,000 people annually in the U.S., often requiring skin grafts for treatment. Recovery involves wearing wound dressings covered by bandage wraps for at least two weeks. While wraps are breathable, versatile, and simple, they can be painful to apply, especially for larger patients, and their compression varies based on the person applying them. This poses challenges when untrained caregivers are involved. Additionally, wraps often slip during physical therapy. Burn care units seek a new solution that matches current wraps in breathability and comfort but offers quicker application, controlled compression, and …