具工單等級之液晶注入排程問題研究

Abstract

在薄膜電晶體液晶顯示器製造過程中，由於液晶注入需要時間較長，為提高其效率採用批次加工生產方式，同時採用平行機台加工，可同時間加工多個批次工件，為滿足緊急訂單、突發狀況等需求，會給予工單不同產品等級，如超級緊急、緊急或一般等不同等級之訂單，具有工單等級之液晶注入排程問題為一個產品族不相容之平行批次加工機台排程問題，以加工機台之總工作負荷最小為目標，同時滿足不同的工單等級、不同之產品族加工時間、不同的到臨時間、順序相依之整備時間、可等候時間有限、批次容量有限等限制，在本論文研究之排程問題中，每一批次之工單等級之決定是由該批次之最高工單等級為該批次之等級，由於此排程問題尚未被研究過，本論文首先建構一混合整數規劃模式來求得此排程問題之最佳解，但若解題時間為重要的考量，則本論文發展兩種演化機制之離散型粒子群最佳化演算法，由於粒子群最佳化演算法同時考量區域搜尋及整體搜尋兩種觀念，本論文所建構之求解模式必定能協助實務工廠的排程運作。

In the thin film transistor liquid crystal display (TFT-LCD) manufacturing factories, liquid crystal injection need a lot of time. To increase the efficiency, parallel machine manufacture is used. To meet the customers’ requirements for rush orders or the development of new product and pilot run for validation test, jobs are always assigned various processing priorities. Super hot lot, hot lot, and normal lot are commonly used processing priorities in factory applications. In this thesis, we consider a new liquid crystal injection scheduling problem with processing priorities (LCISP with priority) which is an incompatible product family batch scheduling problem. The LCISP with priority we investigated in the thesis considered the practical constraints of job processing priority, sequence dependent setup times, unequal ready time, and waiting time simultaneously with the criterion of minimal total workload. It is noted that the processing priorities are dependent on the highest processing priority of all the jobs in the batch. In the thesis, we provide a mixed linear programming model to obtain the exact solutions. If the computational time is a primary concern, two different discrete particle swarm optimization (DPSO) algorithms are also presented by using the merits of local search and global search. It can be said that the solution procedures we provided in the thesis could help the practitioners to obtain better scheduling performance.