The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title:	An application of neural networks to statistical process control
Author:	Papaikonomou, Asterios
View Online:	njit-etd1993-128 (xi, 116 pages ~ 3.1 MB pdf)
Department:	Manufacturing Engineering Division
Degree:	Master of Science
Program:	Manufacturing Systems Engineering
Document Type:	Thesis
Advisory Committee:	Levy, Nouri (Committee chair) Sodhi, R. S. (Committee member) Das, Sanchoy K. (Committee member)
Date:	1993-10
Keywords:	Process control -- Statistical methods Quality control -- Statistical methods Stochastic processes Neural networks (Computer science) -- Industrial applications
Availability:	Unrestricted
Abstract:	Recent changes in the past World War II economy have produced a more educated, value driven customer. A global economy has emerged to technological advances which has made competition fierce as well as proximate. As a result, manufacturing companies are required to provide consumers with quality products at a reasonable cost. A quality product is defined as one that fulfills the needs and expectations of the consumer and provides him with a sense of value. Since all companies want to provide quality products, the reason why poor quality is sometimes the result is not the outcome of actions for that purpose but is due to variability which is inherent in manufacturing processess. A technique which avoids products that don't meet specifications to be shipped out is that 100% inspection is not required and thus the system is inexpensive to implement. The purpose of this thesis is to emphasize the importance of quality, to analyze various aspects of Statistical Process Control and stochastically reproduce some of the nonrandom behavior that often are observed in the industry. Finally, this thesis will explain the development of these expert systems who recognize nonrandom behavior, and recommend methods to continue the research.