Due to the rapid advancement in measurement and sensing technology, large amount of data are commonly generated in various engineering applications. These data contain detailed information of the engineering process, and provide great opportunity for quality improvement. However, the massive data collected from engineering processes usually have complex dependency structure, are influenced by multiple sources of errors and require high computational effort for real time analysis. This thesis contains four essays which propose methodologies to address these issues. In the first essay, a method is proposed to model and monitor surface data with complex local variations. The second essay is concerned with modeling and improving the shape deviation of additive manufactured products, which is influenced by multiple error sources. The third essay proposes a novel method of spatial adaptive sampling and monitoring of high dimensional data streams for detecting clustered out-of-control patterns under the constraint that only a portion of observations can be made at each time, due to the hardware limitations. The fourth essay propose a modeling and predicting method for offline-experimental data with multiple functional responses. These essays provide effective solutions for both offline analysis of experimental data and online monitoring of real-time data with certain complexity that appear in modern engineering applications.
| Date of Award | 2016 |
|---|
| Original language | English |
|---|
| Awarding Institution | - The Hong Kong University of Science and Technology
|
|---|
Statistical methods for quality improvements in data-rich engineering applications
WANG, A. (Author). 2016
Student thesis: Doctoral thesis