Matching Items (2)
Filtering by
- Genre: Doctoral Dissertation
Description
Aviation is a complicated field that involves a wide range of operations, from commercial airline flights to Unmanned Aerial Systems (UAS). Planning and scheduling are essential components in the aviation industry that play a significant role in ensuring safe and efficient operations. Reinforcement Learning (RL) has received increasing attention in recent years due to its capability to enable autonomous decision-making. To investigate the potential advantages and effectiveness of RL in aviation planning and scheduling, three topics are explored in-depth, including obstacle avoidance, task-oriented path planning, and maintenance scheduling. A dynamic and probabilistic airspace reservation concept, called Dynamic Anisotropic (DA) bound, is first developed for UAS, which can be added around the UAS as the separation requirement. A model based on Q-leaning is proposed to integrate DA bound with path planning for obstacle avoidance. Moreover, A deep reinforcement learning algorithm based on Proximal Policy Optimization (PPO) is proposed to guide the UAS to destinations while avoiding obstacles through continuous control. Results from case studies demonstrate that the proposed model can provide accurate and robust guidance and resolve conflict with a success rate of over 99%. Next, the single-UAS path planning problem is extended to a multi-agent system where agents aim to accomplish their own complex tasks. These tasks involve non-Markovian reward functions and can be specified using reward machines. Both cooperative and competitive environments are explored. Decentralized Graph-based reinforcement learning using Reward Machines (DGRM) is proposed to improve computational efficiency for maximizing the global reward in a graph-based Markov Decision Process (MDP). Q-learning with Reward Machines for Stochastic Games (QRM-SG) is developed to learn the best-response strategy for each agent in a competitive environment. Furthermore, maintenance scheduling is investigated. The purpose is to minimize the system maintenance cost while ensuring compliance with reliability requirements. Maintenance scheduling is formulated as an MDP and determines when and what maintenance operations to conduct. A Linear Programming-enhanced RollouT (LPRT) method is developed to solve both constrained deterministic and stochastic maintenance scheduling with an infinite horizon. LPRT categorizes components according to their health condition and makes decisions for each category.
ContributorsHu, Jueming (Author) / Liu, Yongming YL (Thesis advisor) / Yan, Hao HY (Committee member) / Lee, Hyunglae HL (Committee member) / Zhang, Wenlong WZ (Committee member) / Xu, Zhe ZX (Committee member) / Arizona State University (Publisher)
Created2023
Description
This dissertation focuses on a comprehensive exploration of machine learning (ML) and topological data analysis (TDA) with applications for engineering and clinical diagnostics and prognostics. The interface of TDA and ML is called topological machine learning (TML). The key focus and benefit of the proposed TML are on the automated, consistent, and robust handling of high-dimensional data, specifically for the complexities inherent in spatial-temporal datasets. TML's unique ability to capture and quantify high-dimensional geometric and topological features (such as homology) facilitates a deep understanding of the underlying structures of data. The associated dimension reduction capabilities significantly enhance diagnostics and prognostics accuracy and interpretability. TML is first demonstrated using an unsupervised learning setting, where the label information is not required for machine learning. Spatial-temporal data from resting-state functional magnetic resonance imaging (rs-fMRI) are collected and analyzed for Parkinson's disease. Fractal analysis is used to extract topological characteristics of the signal, and extracted features are used in a manifold embedding and projection model for low-dimensional space visualization. The low-dimensional data is integrated with a neural network-based classifier for disease diagnosis. A similar methodology is extended to structural health monitoring problems in engineering. Following this, the TML is developed for a supervised learning setting, where the major application is regression and prediction. Euler characteristics using filtration are used as the topological feature extraction method and extracted features are used in Gaussian Process (GP) modeling for regression analysis. The methodology is first demonstrated with a toy random field problem where a time-dependent field is characterized by varying topological features. The developed method is then demonstrated with crack growth problems with numerical and experimental data. Finally, the topological data analysis is Reflecting on the significant strides made in pushing the envelope of theoretical knowledge while showcasing tangible applications, this work not only charts a course for future progress in the field but also enriches our understanding of machine learning, structural health monitoring, predictive modeling, and beyond. The exploration initiated in this dissertation is just the beginning, with each chapter paving the way for new realms of exploration, innovation, and discovery.
ContributorsXu, Nan (Author) / Liu, Yongming YL (Thesis advisor) / Hong, Qijun QH (Committee member) / Li, Lin LL (Committee member) / Xu, Zhe ZX (Committee member) / Yan, Hao HY (Committee member) / Zhuang, Houlong HZ (Committee member) / Arizona State University (Publisher)
Created2024