本研究以半導體封裝廠為例,探討混合流線型生產環境的批量流與派工法則之排程問題。在已經訂單資訊情況下,考慮各站中不同機群之非等效平行機台與相同機群內完全相同平行機台;各訂單之產品依照規格不同有其適用之機台群限制;各機台加工時間具有隨機現象;加工過程中有其載具轉換所造成之特殊生產特性:黏晶粒站加工之拆批行為決策、模壓站前之集批特性,以最小化訂單之平均流程時間(F ̅)為目標,求得最合適之規劃決策。 本研究面臨之規劃問題主要為三大部分:一為工作站之派工法則選定、二為訂單拆批決策、三為訂單機群指派。由於產業特殊之生產環境問題,且加工時間具有隨機性情況下,其指派問題相當困難且複雜,因此本研究運用模擬最佳化手法進行問題求解,利用基因演算法進行解空間之搜尋與改良法最佳運算資源分配(EGOCBA, Elite Group Optimal Computing Budget Allocation)有效地分配模擬資源以節省模擬時間,求得在同時考量派工法則與批量流之生產情況更能有效降低平流程時間且將方案結果提供給半導體封裝業參考。
This study is a case for Semiconductor Assembly Factory and consider hybrid flow shop problem for lot streaming and dispatching rule. Consider orders with a known arrival time, unrelated parallel machines for machine group, identical parallel machines within machine group, machine eligibility about different product, the processing time with randomness, specific production about lot streaming and batch processor to a scheduling problem. The objective function is to minimize mean flow time of the orders in the study. The problem is to determine the dispatching rule of each stage, the lot size of each job after the first stage, and the assignment machine group of each job to process in each stage. Because of specific production environment, the assignment problem is completive. To address this problem, we develop the simulation optimization approach. To overcome too many alternatives to exhaust, genetic algorithm is used when the search space is large. Elite Group Optimal computing budget allocation(EGOCBA) is used to reduce simulation budget and time while processing time of machines has randomness. The conclusion of this study presents the superior mean flow time in the problem of dispatching rule with lot streaming.