Files in this item



application/pdf9624407.pdf (5MB)Restricted to U of Illinois
(no description provided)PDF


Title:Structural analysis and control of flexible manufacturing systems
Author(s):Lawley, Mark Alan
Doctoral Committee Chair(s):Ferreira, Placid M.
Department / Program:Engineering, Industrial
Discipline:Engineering, Industrial
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Engineering, Industrial
Abstract:This work develops the theory of FMS structural analysis and control. Structural analysis in discrete event systems refers to the characterization of states and regions in the system state space which are logically inconsistent with "normal" system operation. The structural characteristics of an FMS strongly affect its behavior and performance. The primary FMS structural concern is deadlock. Deadlock is the situation in which there is a set of jobs where each job in the set is awaiting buffer capacity occupied by another job in the set.
The state space of the FMS is nicely represented by a directed graph with vertices representing system states and directed edges representing state transition. The objective of FMS structural analysis is to characterize strong regions of this state space which contain the empty state. Such regions are "safe" since strong connectedness with the empty state guarantees that all current jobs can be completed and the FMS emptied. Structural Control Policies (SCP) are real time operating policies that constrain FMS operation to the strongly connected regions.
Ideally, an SCP would admit a state if and only if the state were safe. However, the state safety problem is NP-complete in this context. Thus, accepting every safe state and rejecting every unsafe state will generally be computationally intractable. SCP rejection of all unsafe states in real (polynomial) time implies that some safe states will also be rejected. However, an SCP which rejects too many safe states will seriously impede FMS operation. Also, SCP's must not induce artificial deadlock, i.e. every admissible state must have an admissible safe sequence. Polynomial SCP's which reject all unsafe states and do not induce deadlock are referred to as correct and scalable.
In this work, two correct and scalable SCP's are developed. These policies reject states which exhibit predetermined necessary conditions for deadlock. Procedures for optimizing these policies for specific FMS configurations are presented. A sampling procedure is developed which is used to collect samples of safe states from FMS simulation models. These samples are used to estimate the efficiency (ratio of safe state rejection) of the two policies. Finally, a methodology for dynamically changing routes while maintaining SCP requirements is developed.
Issue Date:1995
Rights Information:Copyright 1995 Lawley, Mark Alan
Date Available in IDEALS:2011-05-07
Identifier in Online Catalog:AAI9624407
OCLC Identifier:(UMI)AAI9624407

This item appears in the following Collection(s)

Item Statistics