Dynamic Network Modeling and Analysis of Large-scale Internet of Things with Manufacturing and Healthcare Applications
Open Access
- Author:
- Kan, Chen
- Graduate Program:
- Industrial Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- December 13, 2017
- Committee Members:
- Hui Yang, Dissertation Advisor/Co-Advisor
Hui Yang, Committee Chair/Co-Chair
Soundar Kumara, Committee Member
Jingjing Li, Committee Member
Lin Lin, Outside Member - Keywords:
- Internet of Things
Process monitoring
Dynamic network
Cardiovascular diseases
Parallel computing
Control chart
Statistical process control
ultraprecision machining
Medical automation
Smart and connected health - Abstract:
- With the advancement of sensing and information technology, sensor networks and imaging devices have been increasingly used in manufacturing and healthcare to improve information visibility and enhance operational quality and integrity. Furthermore, the Internet of Things (IoT) technology has been integrated for the continuous monitoring of machine and patient conditions. As such, large amounts of data are generated with high dimensionality and complex structures. This provides an unprecedented opportunity to realize smart automated systems such as smart manufacturing and connected health care. However, it also raises new challenges in data analysis and decision making. Realizing the full potential of the data-rich environment calls for the development of new methodologies for data-driven information processing, modeling, and optimization. This dissertation develops new methods and tools that enable 1) better handling of large amounts of multi-channel signals and imaging data generated from advanced sensing systems in manufacturing and healthcare settings, 2) effective extraction of information pertinent to system dynamics from the complex data, and 3) efficient use of acquired knowledge for performance optimization and system improvement. The accomplishments include: 1) Fusion and analysis of multi-channel signals. In Chapter 2, a spatiotemporal warping approach was developed to characterize the dissimilarity among 3-lead functional recordings. A network was then optimally constructed based on the dissimilarity and network features were extracted for the identification of different types of diseases. 2) Statistical process control based on time-varying images. In Chapter 3, a stream of time-varying images were represented as a dynamic network. Then, community structure of the network was characterized and community statistics were extracted from time to time. Finally, a new control charting approach was developed for in-situ monitoring of manufacturing processes. 3) Monitoring and Control of large-scale Industrial Internet of Things. In Chapter 4, a stochastic learning approach was developed to construct a large-scale dynamic network from IIoT machines. A parallel computing scheme was further proposed to harness multiple computation resources to increase the efficiency. The constructed large-scale network can be used to visually and analytically monitor machine condition, inspect product quality, and optimize manufacturing processes. 4) Modeling and Analysis of large-scale Internet of Health Things. In Chapter 5, the large-scale network model was extended to the Internet of Health Things. A new IoT technology specific to the heart disease, namely, the Internet of Hearts, was proposed. Dynamic network modeling and parallel computing schemes were designed and developed for multi-level cardiac monitoring: patient-to-patient variations in the population level and beat-to-beat dynamics in the individual patient level. Control charting schemes were further proposed to harness network features for change detection in the cardiac processes.