The full text of this item is not available at this time because the student has placed this item under an embargo for a period of time. The Libraries are not authorized to provide a copy of this work during the embargo period, even for Texas A&M users with NetID.
Modeling and Analysis of Multi-Stage Functional Data for Decision Making in Surface Polishing Processes
Abstract
Surface polishing is a multi-stage process. Different abrasive tools and process parameters are employed at different stages. Tool change and endpoint decisions currently rely on practitioners’ subjective inspections and simple metrics of surface quality. Such practice delivers inconsistent polishing outcomes and delays polishing termination. An automated and principled decision process is pressingly in need. Two types of data, the product data (i.e., surface measurements) and process data (i.e., sensing signals), are available in the polishing processes. By making use of them in different ways, three decision-making schemes are designed for various polishing experiments.
The first scheme is a model-guided criterion developed for flat surface polishing processes. A series of Gaussian process models are trained on the product data, each established for a polishing stage. The models capture the surface variation by a correlation quantifier. A low correlation value reveals the existence of extreme roughness that may be deemed as surface defects. Based on this insight, a decision protocol is designed to enable timely ending actions.
The second scheme is devised for non-flat surface polishing processes. The solution is based on the hypothesis tests with the surface data in curves obtained from two consecutive polishing stages. The test results inform the decision makers whether the current polishing action induces an improvement in surface quality. A significant improvement confirms the effectiveness of the current action and encourages its continuation, while an insignificant or no improvement suggests a tool change or an endpoint.
Unlike the first two schemes where the product data is used, the third scheme uses the process data for polishing decision-making. An offline model and an online detection method are developed. The offline study shows the presence of saturated polishing progression, and the online detection enables breaking a polishing cycle in real time.
When appropriately implemented, the three decision-making schemes in this dissertation could lead to substantial savings in polishing time and energy and significantly improve the throughput of such polishing processes without inadvertently affecting the quality of the final polish.
Subject
Correlation parametersendpoint
Gaussian process
pad change
polishing process
hypothesis test
functional data
inequality
mean curve
variance curve
permutation
change detection
energy turning point
functional features
functional linear regression
saturated polishing effect
surface roughness prediction
interpretable AI
Citation
Jin, Shilan (2022). Modeling and Analysis of Multi-Stage Functional Data for Decision Making in Surface Polishing Processes. Doctoral dissertation, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /198495.