Stocker job scheduling for TFT-LCD plants with evolution strategies algorithm

[[abstract]]完整的液晶平面顯示器製造可以區分為三段，分別為前段的陣列(array)製程、中段的面板(cell)製程、及後段的模組(module)製程。其中，陣列製程由於機台昂貴，最重視的是機台使用率及產能。因此，如何配置這些造價昂貴的生產機台及Stocker設備，並依照其生產流程對機台做最有效的規劃利用，提升面板廠產能對未來生產效率將有著重大的影響。 本研究是針對陣列製程的工廠內的一座Stocker建構模型，利用AutoMod 12 0模擬軟體建構出與面板廠Stocker系統相似的自動化搬運模擬系統模型，並提出相關假設，接著提出搬運策略為指令屬性的線性組合。本研究存取貨指令以先進先出、最佳化求解為主要原則。由於屬性之間的組合複雜且互相牽制，故利用類基因演算法找出適合的搬運策略，如此一來就可以藉由此搬運策略提昇區域產能。經由實驗的結果，在不同投料時間下，最佳化求解模型皆會比傳統的先進先出模型表現的好，更能有效減少搬運時間提昇區域產能。[[abstract]]Thin-Film Transistor Liquid-Crystal Display (TFT-LCD) plants contains three major processes: Array Cell and Module Because the manufacturing equipment in Array plants is more expensive than other plant companies are concerned about the throughput and utilization of the equipment This study models the Array production line and determines the job selection sequence of the stocker to improve the efficiency of the stocker especially during the heavy-loaded production periods We use AutoMod 12 0 simulation software to simulate an automated material handling system (AMHS) model Then we propose relate assumptions and transport strategy which is the linear combination of commands’ attributes In this study Optimization and First In First Out (FIFO) are the two principle transport commands Because the commands’ attribute are more complex and interlinking we use evolution strategies algorithm to find the best transport strategy Therefore we can use this transport strategy to increase capacities Our experimental results show that the performance of optimization model is better than FIFO model in different experiment conditions however we can reduce transporting time and increase throughput effectivel