Date of Award
Master of Science in Engineering
Industrial and Entrepreneurial Engineering and Engineering Management
Industrial and Manufacturing Engineering
Dr. Steven Butt
Dr. Paul Engelmann
Dr. Daniel Mihalko
Masters Thesis-Open Access
There have been many advancements that share similar tools and techniques that help reduce the manufacture of nonconformities. These include computer-aided analysis, design reviews, total quality management, multivariate analysis, process monitoring and control, and root cause analysis to mention a few.
This work details the methodology developed for manufacturing companies to predict attribute defects. Injection molding was used to demonstrate the proposed methodology. Data were collected on a variety of tool design and construction attributes thought to affect the performance of a tool. The independent variables consisted of categorical and numerical data types. The dependent variable was a nominal four-tuple describing the types of defects that can coexist on one part.
A series of steps taken to prepare the data set for classification tree analysis can be categorized by the following: 1) variable screening and selection due to missing data and high dimensionality and 2) causal analysis and similarity computations for combining defects, thus reducing the number of classes in the four-tuple. A method was designed for classification tree analysis. The models provided a way for designers and engineers to assess the potential for success prior to production.
Trahan, Jason S., "A Method for Classification Analysis on Simultaneous Product Defects" (2005). Masters Theses. 4838.