Knowledge-based simulation for initial learning of a dynamic rescheduling system for semiconductor manufacturing

Kathleen Hennessey, Zulfigar Rashid, Kwang Soo Hahn, Milton Smith

Research output: Contribution to journalConference articlepeer-review

Abstract

Simulation techniques provide the initial knowledge base for an artificial intelligence scheduling system. Such an approach reduces the initial learning period, during which time the AI system is not productive, and avoids the time-consuming and error-prone process of rule preparation by one or more experts. Threshold logic, a technique associated with neural nets, is the main learning mechanism for dynamic rescheduling of semiconductor manufacturing processes. A discrete next-event simulation technique simulates a schedule for the IC manufacturing system. Whenever a new event is created, it is stored in an event list that is sorted in order of increasing time. There are two types of events: lot events and machine events; lot events include the actions of selecting a lot and getting a new lot from the lot source. Information about the system model concerning processes and machines is stored in a file which allows the system configuration any number of machines and processes.

Original languageEnglish
Pages (from-to)var paging 780
JournalTechnical Paper - Society of Manufacturing Engineers. MS
StatePublished - 1989
EventSemiconductor Manufacturing Conference - Phoenix, AZ, USA
Duration: Nov 14 1989Nov 15 1989

Fingerprint Dive into the research topics of 'Knowledge-based simulation for initial learning of a dynamic rescheduling system for semiconductor manufacturing'. Together they form a unique fingerprint.

Cite this