Electronic Theses and Dissertations

Identifier

4943

Author

Ming Fan

Date

2017

Date of Award

4-19-2017

Document Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Engineering

Concentration

Mechanical Engineering

Committee Chair

Gang Qi

Committee Member

Steven F Wayne

Committee Member

Lih Y Deng

Committee Member

John I Hochstein

Abstract

Damage is a form of micro-structural change that diminishes the mechanical and/or physical properties of solid materials. The evaluation of damage is needed to assess the reliability and remaining service life. Prior efforts to characterize damage were driven by physical laws or those driven by data. The former is dominated by searching analytical formulations according to given physical laws, whereas the latter is to mine the massive data as indicators that can be used for the assessment. A two-dimension data matrix DA was developed from acoustic data on metals and polymers, and used as the information infrastructure. The DA matrix was detailed from the points of views of informatics. However, there are a number of unsolved fundamental issues regarding this data matrix, for instance, means to construction of DA such that its rows and columns to be physically meaningful; the influence of construction of DAon the embedded information; and means to construction of DA that counts for multiple experimental results. Furthermore, there is a prominent need for a comprehensive computer program that integrates the construction of DA, information quantification, statistical analysis and visualization of results. The primary goal of this work is to address the above fundamental issues systematically. The means were developed to determine the number of rows (observations) and columns (scales) of DA by 1) direct means and 2) reversed means. Also, basic statistical calculation were introduced into the construction of DA matrix to check the validity of the result of data operation.The results shown that the statistical calculation can help to capture the full potential information from multiple experiments. We also demonstrated the algorithms of data operation through the clustering, classification, correlation, estimation, and prediction and showed the corresponding visualization results. To address the integration of data representation, data operation, and data visualization, a software program, IntelliAE, was developed to construct the information infrastructure and facilitate informatics analysis. IntelliAE is a platform that will deliver other scientific significance and application value on the microstructural change informatics resulting from damage in materials.

Comments

Data is provided by the student.

Library Comment

dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.

Share

COinS