DATA FLOW ARCHITECTURES FOR DISTRIBUTED CONTROL OF COMPUTER OPERATED MANUFACTURING SYSTEMS: STRUCTURE AND SIMULATED APPLICATIONS. (VOLUMES I-IV)
Abstract
The investigation defined and tested a new class of control algorithms for computer operated manufacturing systems (CMS). The new class was to be broadly applicable, to be responsive to batch weight, to give high machine tool utilization without either static scheduling or severe restrictions on part entry, and to be reliable, repairable and extensible. The definition demonstrates the existence and generality of the new class. The definition is sufficient to permit construction of any element of the new class by a practitioner with backgrounds in electronic devices, NC machine tools, computer operating systems, and data flow. Any exisiting and most foreseeable CMS can be controlled by an element of the new class. Reliability, repairability, and extensibility are considered during the definition. The test applied the new class to control two simulated CMS, one like existing systems, the other using adaptive machine tools. The simulation is written to be verifiably correct, and appears with documentation in an appendix. The new class functioned successfully for both systems. The control was of good quality. Non-failed machine tool utilization exceeded 95% for failure rates from 3% to 16% per machine tool. The batch weights had a strong effect on relative flow time. The dissertation concludes that the new class of CMS control algorithms exists, may be applied to a broad class of CMS, and for the cases simulated gives high machine tool utilization and strong response to batch weight. It would probably increase reliability, repairability, and extensibility.
Degree
Ph.D.
Subject Area
Industrial engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.