| 摘要: | 本研究針對流程型、非等效平行機台、機台合適度、製程相依及製程獨立設置時間的生產排程問題進行探討並考量求解目標為 : 最小化加總成本(ΣCj)及加總?對延遲時間。
本研究針對個案工廠實際的生產工單組合進行分析，先?用限制理論把全線的排程問題限縮在前段有製程設置時的三個工作站，再針對這三站的製程特性進行個別分析，再把問題範圍縮小到—非等效平行機台、有機台合適度、產能平衡、設置時間議題的綠漆工作站，為全線的產能瓶頸站，並利用限制驅導式排程及Permutation的觀念，以綠漆站的排程當做投料排程來簡化全線的生產排程問題。再利用Excel內建的啟發式演算法做求解的工具，但因Excel的限制再並需再將α= P2類型的排程先，以人工的方法將工單分成兩類使排程問題又從P2再簡化為1(1*2 , 1R / 1S)。
本研究先建構一個評估個案工廠排程績效的數學模式，運用個案實際的工單資料為測試基礎，再以SPT、EDD、LST三種派工法則及個案現行的人工排程方法與Excel內建的啟發式演算法，做排程績效之評估比較，結果發現Excel內建的啟發式演算法在換線成本項目的績效上有10 ~ 43% 的提昇，但在加總?對延遲時間項目的改善上則不顯著。
This study explores the production scheduling problems of Floor Shop, Unrelated Parallel Machine, Machine Eligibility, Sequence Dependent Setup Time and Sequence Independent Setup Time and also attempts to solve the following research question: the minimization of ΣCj & ΣTj.
This study also analyzes the combinations of run card in the case-study factory through the following procedures. First, confine the problems of the whole workstations scheduling to the three working stations with anterior section of having setup time process by means of TOC. Next, implement the independent analysis of the process feature in these three stations, and then further restrict the problems to a certain Solder musk working station, featuring Unrelated Parallel Machine, Machine Eligibility, Balance of Capacity and Setup Time Issue, which is the bottleneck station in the production line. Then, with the concept of D-B-R, simplify the problems of production scheduling in the whole line by assimilating the scheduling in Solder musk station into goods issue scheduling. Finally, use the Meta-heuristics built in the Excel as the means of instruments.
This study begins with a mathematical module to evaluate the performance of effects in the case-study factory and uses the authentic data of run card as the test foundation. Then, activate the comparison of performance of effects by means of three Dispatching Rules – SPT, EDD, and LST, artificial scheduling, as well as the Meta-heuristics built in the Excel. The result demonstrates 10 to 43 percent of improvement in the item of cost of setup calculated by the Meta-heuristics built in the Excel. However, there is no significant improvement in the item of ΣTj. |
