TFT-LCD產業為台灣現今明星產業之一,正處於起飛階段;科技製程技術不斷精進,為了降低單位成本,因而需要持續不斷地大量投資,研發新製程技術,擴充廠房,所以須靠大量接單生產製造才能弭平成本進而有利可圖。在模組組裝製程中的物料規劃程序,規劃良好與否會直接影響到之後成品的生產數量,因此需要一套能最佳化規劃TFT-LCD物料配置的決策模式,以圖能夠在許多限制之下,使得生產產量盡量最大。現階段TFT-LCD產業乃利用先進規劃系統作為其輔助決策工具,但仍有不足的地方,本研究將以蟻群最佳化演算法(ACO)為基礎來發展一套最佳化物料規劃決策模式,並利用田口法來參與參數設計部分,來使得此模式效益能達到最佳;在實驗結果方面,將會與基因演算法(GA)以及貪婪法則(Greedy)做比較,期盼能有效率地解決TFT-LCD物料規劃與配置方面的問題。
TFT-LCD is now on developing and it is one of the focal industries in Taiwan. Generation technology has advanced all the time. In order to reduce the unit cost, the TFT-LCD industries need to invest much money in inventing new technology, and expanding factory building, therefore it must take orders in mass production to fill up the cost, and then to earn profits. In the material planning process of module process, planning good or not will influence the quantity of finished products. For this reason, we need a decision model to make TFT-LCD material planning optimal which enables the quantity of production more under much restriction. Today, the decision model in the TFT-LCD industries is APS, but it still insufficient. Our research will be based on Ant Colony Optimization algorithm (ACO) to develop a optimal decision model in planning material, and use Taguchi Method in parameters design to make this model better. In the experiment, the computational results on test problems will be in comparison with Genetic Algorithm (GA) and greedy method, and then could solve the problem of material planning and allocation efficiently.