11115 research outputs found
Sort by
A Reinforcement Learning Approach to the 2v2 Beyond Visual Range Air Combat Maneuvering Problem
Get PDFThis research examines a 2v2 air combat maneuvering problem (ACMP) in a Beyond Visual Range (BVR) environment. A discrete-time, infinite-horizon Markov Decision Process (MDP) model represents the BVR-ACMP, seeking to determine high-quality policies for a pair of autonomous aircraft to execute tactical maneuvers and firing decisions. The Advanced Framework for Simulation, Integration, and Modeling (AFSIM) characterizes the complex six-degree of freedom (6-DOF) aircraft operations, encompassing kinematics, sensors, and weapons. Given the high dimensionality and continuous nature of the state and decision variables, a deep reinforcement learning (RL) solution approach is adopted wherein the value function is approximated via a Neural Network (NN). The research includes designing neutral starting state scenarios for training and assessing the impact of adversarial behaviors and missile characteristics on decision policies. A three stage hyperparameter tuning experiment is conducted to obtain high-quality policies. Several case studies are examined to evaluate the effectiveness of the deep RL approach, demonstrating its feasibility for generating aircraft behavior models for air combat AFSIM based simulation studies
Public-Private Partnerships (PPPs) in the Defense-Critical Semiconductor Industry: A History and Firm Level Analysis
Get PDFThis thesis provides 3 contributions to the literature of semiconductor industry industrial policy. The thesis (1) provides a historical overview of industrial policy within the U.S. semiconductor industry, (2) measures semiconductor public-private partnership (PPP) participation and (3) measures firm functions’ impact on firm level performance. PPP participation data was gathered from a sample of 10 semiconductor PPPs. Firm level performance data was collected from 66 public, U.S., SIC 3674 “Semiconductor and Related Devices” firms from 2014 to 2022, and a random effects, fixed effects, and pooled ordinary least squares regression was conducted on an unbalanced panel data set of 524 firm-year observations. We find that (1) PPP participation does not significantly impact 3 firm level performance metrics: ROA, EBITDA margin, or revenue growth; (2) firm function, by design-only and manufacturing firms, does not significantly impact the same 3 firm level performance metrics. Lastly, no statistically significant results support that firm level performance differs by the (3) function of firms that participated in semiconductor PPPs
Assessing Military Parking: A Deep Learning Approach to Evaluating Standards and Impacts
Get PDFCurrent United States Department of Defense (DoD) standards require a minimum amount of parking for each building. This requirement defines how much off-street parking to construct. However, the impact of these requirements remains unclear. This study builds upon the emerging field of overhead imagery analytics by directly tying it to parking on military installations. Specifically, this study leverages a pretrained deep learning car detection model, Car Detection – USA, developed by Esri for use within ArcGIS, and couples it with open-access temporal imagery sourced from Google Earth Pro to assess selected parking lots across Area B, Wright-Patterson Air Force Base, a representative United States military installation, during a 10-year period. This study determined the model is suitable for assessing parking lot usage due to its high level of performance when presented with a test dataset. However, it is poorly suited for such assessments without manual user oversight and validation and the use of high-resolution imagery as it tends to underestimate parking lot usage when presented with a deployment dataset. Additionally, this study determined that parking supply within assessed parking lots exceeded demand across the 10-year assessment period, resulting in unnecessary excess financial burdens in operating and maintenance costs. These results suggest parking policy standards within the DoD may necessitate excessive parking, indicating a broader DoD-wide issue. To this author\u27s knowledge, this research is the first to combine deep learning object detection software with satellite imagery data to study parking policy, parking lot usage, and its economic impacts on a representative DoD installation
Computational Analysis of Transverse Jet Interactions on a Cone In Mach 3.9 Crossflow
Get PDFThe objective of this investigation was to analyze jet interactions using numerical simulations, which was compared to experimental data. A sharp cone with a 7-degree half-angle was modeled, along with the wind tunnel, using a 1:1 scale. Analysis of the data found the flow characteristics of the jet, including the bow shock and recirculation regions, were captured nicely in the CFD simulations, as well as the oblique shock off the nose of the cone. CFD simulations of the cone in a freestream environment depict shock-shock interactions not captured by experimental schlieren images. This research provides a rare glimpse into CFD simulations of the wind tunnel, while also offering insight into the jet interaction effects not easily obtained using experimental methods
Advanced Oxidation of Methyl Tert Butyl Ether with Ultraviolet Light Emitting Diodes
Get PDFMethyl Tert-Butyl Ether (MTBE) is a volatile, soluble, organic compound introduced to gasoline within the United States as a fuel oxygenator in 1979. MTBE is a common groundwater pollutant due to its high water solubility and resistance to bio-degradation; it commonly infiltrates groundwater sources via leaking underground storage tanks or fuel spills. This research employed ultraviolet light emitting diodes (UV-LEDs) and hydrogen peroxide (H2O2) induced advanced oxidation process (AOP) to treat MTBE contaminated water at the bench scale. Using four H2O2:MTBE molar peroxide ratios (100:1, 200:1, 400:1, 500:1), MTBE degradation was induced in a continuously stirred tank reactor (CSTR) and quantitively analyzed via Gas Chromatography-Mass Spectrometry (GCMS). Using calculated pseudo-first-order rate constants (ks) as a performance metric, the molar peroxide ratios of 200:1 and 500:1 resulted in high levels of MTBE degradation, producing ks values of 0.0460 min1 and 0.0403 min1, respectively. Statistical analysis determined the optimal molar ratio to be 200:1 (average ks=0.0460 min1) due to no statistical difference between the 200:1 and 500:1 molar ratios, resulting in up to 50% MTBE degradation. Potential degradation pathways and byproducts were theoretically evaluated using Density Functional Theory (DFT). Tert-butyl Alcohol (TBA), Tert-butyl Formate, Methanol, and Acetate were designated potential degradation byproducts, with TBA acting as a potential hydroxyl scavenger and limiting MTBE degradation. The results contained within this research will serve as a baseline for utilizing AOP for groundwater pollutant remediation
End-User Device Management In The Air And Space Forces
Get PDFWhen achieving technical superiority is a matter a national security, it is critical that the Department of the Air Force (DAF) employs a robust Information Technology Asset Management (ITAM) strategy that maximizes mission effectiveness. Since 2022, the Office of the Chief Information Officer (SAF/CN) has been developing a plan to transform the Air Force’s ITAM strategy, primarily intending to centralize End User Device (EUD) procurement to the DAF level. For this study, a Delphi study was conducted to gather expert opinions on key aspects of SAF/CN’s new ITAM strategy to identify strengths, weaknesses, and implementation challenges. This research provides valuable insights and recommendations to SAF/CN and relevant stakeholders. There were several key findings as a result pf this study. Experts identified advantages and disadvantages that centralized ITAM would likely experience. Overall, experts recommended that the DAF strike a balance between centralized and decentralized approaches to maximize the benefits of each
Small Unmanned Aircraft System Detection and Tracking with Audio, Computer Vision, and Deep Learning Techniques
Get PDFsUAS present significant risks to local and federal agencies when under the control of negligent, reckless, or criminal operators. In the face of an escalating presence of sUAS in shared airspace with traditional aircraft, and their deployment in protected airspace as potential weapons, safeguarding personnel, facilities, and assets becomes paramount. This research seeks to address this emerging threat by investigating the efficacy of integrating low-cost distributed sensors and Machine learning (ML) models to enhance battlespace awareness and complement existing sensing platforms for real-time sUAS detection, classification, and localization. The thesis introduces the conceptualization and development of a Drone Detection Command Center (DDCC). The DDCC interfaces with a distributed node system de- signed to acquire sUAS data in real-time through both visual and audio modalities. Furthermore, the DDCC possesses the ability to capture frames of interest, facilitating their integration into future machine learning models for enhanced prediction capabilities. Data is collected of the DJI Matrice 600 Pro using the system and the data is used to create multiple deep learning models with the goal of classifying sUAS presence and predicting the range the sUAS is from a given node. A focus is placed on evaluating the performance of range predictions based on audio, then comparing those to the range predictions based on video. Finally, the data is fused into a single dataset, and the same predictions are made on a custom model with the goal of determining if fused data will present superior results to individual modalities. As an initial step, audio classification achieves a categorical accuracy of 79.6%, while video classification achieves an accuracy of 86.7%. From there, range predictions are iv made with audio and video datasets independently, providing a mean absolute error of 10.463 meters for audio, and 16.961 meters for video respectively. Finally, the audio and video data is fused together and fed into a Convolutional Recurrent Neural network to test if the combined data will provide better results. The combined results show that the mean absolute error is 9.57 meters, an improvement of .88 meters over audio data, and 7.385 meters over video data
Trajectory Optimization via Direct Collocation Methods of Boost-Glide Vehicle with Newtonian Aerodynamic Model
Get PDFIn this work, a Newtonian aerodynamic model was developed for an elliptic cone glide geometry at a variety of angles of attack, and a functional relationship between an input angle of attack and the vehicle lift and drag coefficients was developed. This model was used in a direct orthogonal collocation framework to simulate a series of boost-glide trajectories, with objectives of minimum distance to target, minimum time to target, and maximum velocity to target. The optimal control problem was then expanded into a reachability study, considering a wide range of additional scenarios, along with studies on the vehicle range capabilities. Results demonstrated a robust framework for creation of hypersonic trajectories, and the ability to implement a rapidly generated aerodynamic model to capture higher-fidelity behavior in the simulation
Positioning, Navigation, and Timing (PNT) Study for Select Multibody Environments
Get PDFIn the renewed 21st-century race to the Moon, the demand for heightened mission capabilities in the cislunar region, lunar surface, and beyond is imperative. With an escalating number and complexity of lunar missions, persistent positioning, navigation, and timing (PNT) capabilities are becoming paramount. This thesis presents innovative lunar PNT architectures designed to provide continuous coverage of the lunar South Pole, near-continuous coverage of the entire lunar surface, and coverage of the Earth-Moon Corridor. As space faring countries advance from lunar exploration to Mars, similar assets are essential for astronaut exploration in this extraterrestrial environment. Additionally, this thesis introduces and meticulously analyzes several constellations for Mars and Phobos PNT. The research delves into orbits propagated in the Circular Restricted Three Body Problem (CR3BP) and Bi-Circular Restricted Four Body Problem (BCR4BP) through methods such as initial condition propagation and Poincaré Mapping. A comprehensive performance analysis is conducted, considering factors such as visibility coverage, system power considerations, position dilution of precision, geometric dilution of precision, and stability in both three-body and four-body systems. These analyses are crucial for narrowing down the options and identifying the most fitting architectures. For South Pole coverage, the study favors a constellation of Elliptical Lunar Frozen Orbits (ELFOs) due to its consistent coverage of the lunar South Pole. In the case of entire lunar surface coverage, two constellations of ELFOs and Walker Deltas emerge as equally viable options. Earth-Moon coverage, deemed challenging due to the vastness of space, leans towards a 3:1 Resonant Orbit with a Distant Retrograde Orbit (DRO) and a 2.5 xGEO orbit as a potentially optimal constellation. Moving beyond the Moon, an Axial constellation surpasses expectations for Martian PNT. For Phobos PNT, the architecture design falls back on a Walker Delta constellation. The thesis concludes with numerous findings and recommendations, including a comprehensive exploration in the appendix that sheds light on the profound impact of optimizing satellite placement on results
