Integration of simulation and neural network in forecasting the throughput for TFT-LCD colour filter fabs

Abstract

In manufacturing industry, the strategy of operations management influences the performance of production. The work in process control and order release policy are two important factors that would affect a decision-making of strategy. To evaluate the performance of these factors, simulation has been widely applied in recent years. Since running simulation always takes time, it is difficult to provide information promptly for production planners to respond to practical situation in industry. To solve this issue, a simulation assisted neural network forecasting system of throughput for thin film transistor liquid crystal display colour filter fabs is developed. This system enables planners to conduct what-if analysis for the production control policies without disturbing operations in a real plant. The obtained results show the forecasting system has the same capacity to estimate the throughput of production as a simulation model, but the response time is much faster.

Keywords:

• colour filter fabs
• neural networks
• simulation
• throughput
• CONWIP
• order release

ORCID

PoTsang B. Huang http://orcid.org/0000-0003-4050-6885

Additional information

Funding

This research was supported by National Science Council of Taiwan, ROC [contract no. 100-2622-E-007-022-CC2] and Chunghwa Picture Tubes Ltd. Taiwan ROC.