本研究利用基因演算法求解 TFT-LCD 產業之陣列廠設施規劃。以最小化玻璃基板的總平均搬運距離為目標函數。透過基因演算法軟體 Evolver 進行模式建構,針對四種產品比例提出其具有最短搬運距離條件之最適佈置,並比較各個佈置在四種不同產品比例下之表現。實驗結果顯示,由產品比例 1:1:1 所得到的設施佈置方案,其平均單片的移動距離優於其他方案。楊家霖 (2008) 提出一套步驟式的設施規劃程序 (Fab Design Procedure, FDP),能快速發展與評估 TFT-LCD 陣列廠建廠初始設計方案,本研究建構之基因演算法模式於建立設施佈置之表現優於 FDP。
Genetic algorithm (GA) is used to solve the facility planning problems of TFT-LCD array fab in this study. It minimizes glass substrate’s average moving distance as the objective function. Research model is formed by Evolver. The model produces four optimal layouts with the shortest distance from the arrangement of transport from four types of product mix, and it yields the best performance across four scenarios in terms of the distance. Results indicate that the layout of product mix 1:1:1 is better than other layouts in shortening the average sheet moving distance. The proposed GA method outperformed the Fab Design Procedure (FDP), a step-by-step approach to conduct quick calculations to develop and evaluate the initial design alternatives for TFT-LCD array fab, proposed by Yang in 2008.