Multi-criteria design and control of manufacturing systems using simulation and artificial intelligence
Abstract
The objective of this research was to develop a prescriptive simulation system for manufacturing: the current systems have been found to suffer from short-sightedness (myopia) and are unable to satisfy multiple performance criteria simultaneously. To ameliorate the deficiencies of current systems, an autonomous prescriptive simulation system (APSS) is developed based on the opportunistic model of problem-solving. The problem of myopia in prescriptive systems is defined and a cure proposed--based on knowledge of the structure and behavior of the problem domain. System interference in manufacturing is quantified and analyzed, and the results used in the development of the prescriptive system. A conflict-resolution algorithm is developed for multi-criteria design and control of manufacturing systems--based on goal-regressing planners used in the field of artificial intelligence. The system is evaluated against a discrete-valued stochastic optimization technique and found to perform satisfactorily. The system outperforms the optimization technique in all cases where the defining variable set is large. Finally, it is noted that a two-level architecture for a prescriptive system based on domain knowledge and blind-search may provide better results than one based on domain knowledge alone.
Degree
Ph.D.
Advisors
Talavage, Purdue University.
Subject Area
Industrial engineering|Artificial intelligence|Operations research
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.