The operation of large computer controlled manufacturing systems
Abstract
This work examines methods for operation of large computer-controlled manufacturing systems, with more than 50 or so disparate CNC machines in congregation. The central theme is the development of a distributed control system, which requires minimal central supervision, and allows manufacturing system re-configuration without extensive control software re-writes. Provision is made for machines to learn from their experience and provide estimates of the time necessary to effect various tasks. Routing is opportunistic, with varying degrees of myopia depending on the prevailing situation. Necessary curtailments of opportunism are built in to the system, in order to provide a society of machines which operate in unison rather than in chaos. Negotiation and contention resolution are carried out using a UHF radio communications network, along with processing capability on both pallets and tools. Graceful and robust error recovery is facilitated by ensuring adequate pessimistic consideration of failure modes at each stage in the scheme. Theoretical models are developed and an examination is made of fundamental characteristics of auction-based scheduling methods. By means of simulation study, the tolerance of the system to failures in communications and performance under various policies are explored. The effect of reductions in system flexibility is examined, this quality having been quantified by the development of an information-theoretic metric.
Degree
Ph.D.
Advisors
Barash, Purdue University.
Subject Area
Industrial engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.