半導體製造產業的生產和服務系統中,批量生產方式已成為生產的主要工具之一,生產系統定必以「多樣小量」的形態為主流,而時間變數對於生產系統的控制,也是一個不可忽視的因素。 本研究以機台的兩種裝載方略 (Machine Loading Policy) 之機台設定機率模型 (Time Batching Policy、Job Batching Policy) 作為研究對象。兩種方略在不同產品組合之下,其時格大小 (Time Bucket Size) 與批量大小 (Batch Size) 如何影響機台設定次數,針對產品多樣性的製造,分析其機台設定機率,並以模擬軟體 (Arena) 設計之模型作為驗證之工具。在分析兩種方略時,發現不均勻的產品組合會得到較低的機台設定機率。及後利用易於時間控制的Time Batching Policy,以其計算出的機台設定機率作為工單對機台分配的依據,發現產品組合只跟工件種類數和到達率有關,與機台分配無關,從而歸納出Meta-Knowledge,免除於工單與機台限制改變時需要重新計算最佳生產排程時所耗費的時間。
In the semiconductor manufacturing industry, Batch mode of production has become one of the main tools for production, the time variable for the control of the production system is important factor. In this study, probability of setup model for two machine loading policy (Time Batching Policy and Job Batching Policy) is studied. Under a Product Mixes, Time Bucket Size and Batch Size how to affect the setup times for the product variety on the manufacturing system, and use simulation model on Arena as a tool for verification. Analysis of both policies and found that uneven product mixes will gain lower probability of setup. Design the Job-Machine Assignment Model based on Time Batching Policy, find that product mixes affected by arrival rate, rather than machine assignment. After generalize the Meta-Knowledge, exempt from recalculate the optimal production scheduling when the work orders or machine capability restrict are changed.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。