A connectionist expert system for learning process drift control in offset lithographic printing

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Mechanical and Aerospace Engineering


Young Bai Moon


Mechanical engineering, Artificial intelligence

Subject Categories

Mechanical Engineering


Techniques from artificial intelligence research have provided a foundation for the development of decision-making systems for automation at the knowledge level. By studying the role of the human operator in the on-line control of process drift, this research investigates the nature of the knowledge that is specific to the operator and the machine. The primary objective is to develop a representation that models the relationship between the observable and adjustable variables of a process as perceived by an experienced operator when compensating for process drift. The phenomenon of process drift is studied in the context of offset lithographic printing where it is a prominent problem. The on-line adjustments that a proficient press operator applies to compensate for process drift are specific to a particular machine. The nature of the operator's knowledge was found to consist of articulated relationships between process variables and an unarticulated conceptualization. An integrated connectionist expert system was developed to model the operator's machine-specific knowledge. The articulated relationships were represented in the form of production rules. Whereas, a connectionist system was designed to learn from actual process data generated by the operator during on-line control. It was found that the relation between the subtractive primaries of the ink colors and the additive primaries has been uncovered by the connectionist representation. A weight-based conflict resolution technique was developed to integrate the connectionist representation with the expert system.


Surface provides description only. Full text is available to ProQuest subscribers. Ask your Librarian for assistance.