A learning-based methodology for scheduling problems in semiconductor fabrication plants
Abstract
In the last decade, regardless of whether the world economy was in recession or prosperity, the semiconductor business has continued to propel not only the electronics industry but many other industries as well. However, semiconductor manufacturing is among the most complicated and capital-intensive manufacturing processes in the world. Effectively operating wafer fabrication plants is the greatest challenge and an important task. In this study, a framework to develop learning-based scheduling systems is proposed. The key components of the framework include a learning-biases selection procedure, and a three-level scheduling mechanism. Through the case studies conducted, the proposed bias selection procedure has demonstrated its capability to select appropriate learning biases. In addition, the learning-based scheduling system derived from the framework proposed shows promising results in increasing system throughput and reducing the time required to process semiconductor wafers.
Degree
Ph.D.
Advisors
Yih, Purdue University.
Subject Area
Industrial engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.