Physical-Statistical Modeling and Optimization of Complex Systems - Healthcare and Manufacturing Applications
Open Access
- Author:
- Yao, Bing
- Graduate Program:
- Industrial Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- June 05, 2019
- Committee Members:
- Hui Yang, Dissertation Advisor/Co-Advisor
Hui Yang, Committee Chair/Co-Chair
Soundar Kumara, Committee Member
Eunhye Song, Committee Member
Xingyuan Fang, Outside Member
Soraya M Samii, Special Member - Keywords:
- Physical-statistical modeling and optimization
Spatiotemporal regularization
Sequential decision making
Energy-efficient mobile sensing
Multifractal analysis
Quality control
Additive Manufacturing
Inverse ECG modeling - Abstract:
- The rapid development in sensing and information technology facilitate the effective modeling, monitoring, and control of complex systems. Advanced sensing and imaging have brought a data-rich environment and provided unprecedented opportunities to investigate system dynamics and further optimize decision making for smart health and advanced manufacturing. However, the sensing data is generally with high-dimensionality and complex structures. Realizing full potentials of sensing data depends to a great extent on novel analytical methods and tools with effective information-processing capabilities. The objective of this dissertation is to advance the knowledge on sensor-based system monitoring, modeling, and optimization by developing innovative physical-statistical methods for smart health and advanced manufacturing. This research will enable and assist in 1) handling high-dimensional spatiotemporal data; 2) extracting pertinent information about system dynamics; 3) optimizing decision making under uncertainty. My research accomplishments include: • Energy-efficient mobile ECG sensing: In Chapter 2, an energy-efficient framework is proposed for mobile ECG sensing through the constrained Markov decision process, where the sensing policy is optimized by maximizing the detection accuracy of cardiac events under the constraint of energy budget. • Physical-statistical modeling of space-time complex systems: In Chapter 3, a physics-driven spatiotemporal regularization method is developed for high-dimensional predictive modeling. This model not only captures the physics-based interrelationship between time-varying explanatory and response variables that are distributed in the space, but also addresses the spatial and temporal regularizations to improve the prediction performance. • Spatiotemporal inverse ECG modeling: In Chapter 4, a robust inverse ECG model with spatiotemporal regularization is developed to reconstruct the heart-surface electric potentials from body-surface sensor measurements. Furthermore, a wavelet-clustering method is proposed to investigate the cardiac pathological behaviors from the reconstructed heart signals and characterize the location and extent of myocardial infarctions on the heart surface. • Multifractal analysis for nonlinear pattern characterization: In Chapter 5, a multifractal approach is developed to quantify the nonlinear and nonhomogeneous patterns in image profiles for defects identification and characterization in additive manufacturing (AM). • Sequential optimization and real-time control of additive manufacturing processes: In Chapter 6, a sequential decision-making framework through the Markov decision process is proposed to optimize the AM build quality layer-by-layer. This framework enables on-the-fly assessment of AM build quality and real-time defect mitigation.