A fuzzy-based approach to the liquid crystal injection scheduling problem in a TFT-LCD fab

Abstract

In a thin film transistor-liquid crystal display (TFT-LCD) fab, the liquid crystal injection (LCI) process is to inject liquid crystal into the cell gaps on panels. Since its processing time is enormous (typically 12 h) compared to other processes, the LCI process is a bottleneck in the entire cell process. This study focuses on the LCI scheduling problem, which is divided into two sub-problems: automated guided vehicle (AGV) dispatching and LCI machine scheduling. A self-adjusted fuzzy (SAF) method is developed to solve the AGV dispatching problem. The SAF method is fuzzy based, and it is capable of adjusting the inference rules according to the status of the system to determine which cassette is to be transported first. A modified least slack time (MLST) method is proposed for the LCI machine scheduling problem. The MLST method assigns available LCI machines to first work on processing batches which will be finished beyond their due dates. If there are no such batches, the system releases a new batch, which is waiting in the input buffer with the least slack time, to the available LCI machine. Results indicate that the proposed SAF and MLST methods are able to finish a certain number of batches in a shorter time and reduce the tardiness of cassettes.

Keywords:

• LCI
• scheduling
• AGV dispatching
• fuzzy
• self-adjusted
• TFT-LCD