半導體等高科技產業,為提高昂貴設備之利用率,製造現場往往採取輪班方式,如何排班是屬於NP-hard 性質的問題,如果又遇到旺季需要增加產量的情況,將會使排班變得更耗費時間與人力。過去文獻多以基因演算法與整數規劃來進行研究,比較少針對粒子群演算法(Particle Swarm Optimization, PSO)來解排班問題,同時大多都是針對護理人員的排班進行探討,極少針對半導體產業的排班方式進 行探討。本研究主要目的,乃提出一個以粒子群演算法導入產量績效指標的基礎排班機制,藉以協助解決半導體產業於人員排班的例行性工作。本研究導入產量績效指標之限制條件,並以粒子群演算法來進行求解,發展一快速排班的演算法,以解決排班問題,針對5 人排7 天與20 人排31 天的實驗結果顯示出此本研究排班模式在排班時間、滿足產量等方面皆有不錯的效果,也可提供半導體業者排班人員為排班的參考依據。
n the semiconductor industry, in order to raise the utility of expensive equipment, manufacturing department usually adopt various shift systems. The staff scheduling problem is an NP-hard problem. It will be difficult to arrange and will spend a lot of time and manpower in the busy season needed to increase production.Previous research about staff scheduling usually used genetic algorithms and integer programming. This study proposes a particle swarm optimization (PSO) based scheduling mechanism integrated with performance indicator for semiconductor industry. One performance indicator, production quantity, is considered in the staff scheduling problem. Solved by the PSO, the approximate solution can be found in a short time. Experimental results show the staff scheduling model proposed in this study can quickly obtain the shift schedule and satisfy the production quantity simultaneously. It also offers a reference for semiconductor industry scheduling.