群組製造 (cellular manufacturing) 已普遍運用於高度客製化及複雜化的小至中批量生產,實務上的廣泛運用使得具有順序相依整備時間之流程型製造單元排程問題 (flowline manufacturing cell scheduling problem with sequence-dependent family setup times) 成為近年來排程領域活躍的研究課題之一。本研究擬探討具有訂單拒絕之流程型製造單元排程問題,因在有限產能的限制下,實務上廠商不一定能在要求的期限內完成所有的訂單,若超過最後期限可能會造成顧客的不滿,導致懲罰成本的發生,所以若能拒絕無法順利完成的訂單會增加公司的獲利能力,因此,引發本研究探討具有訂單拒絕之流程型製造單元排程問題並提出和弦演算法,用以求解最大化總收益問題,以使得多樣少量的生產能得到類似大量生產型態的經濟效應。為了驗證此演算法,在小問題上,我們與最佳解法做比較以驗證其求解績效,由實驗結果得知在小問題上其差異不大但隨著工件數增加,原始的和弦的求解品質會越來越差,我們相信此研究可以帶給之後想要繼續往下做的人一點啟發。
The Flowshop manufacturing cell scheduling problem (FMCSP) with sequence-dependent family setup times (SDFSTs) has become a key area in the field of scheduling. We want to discuss the FMCSPs with rejection because the company capacity is limited in the real word. We can not satisfy every customer. Therefore, we need to know how to choose the right orders, because if we can not finish the product before the customer’s deadline, the product will lose its value. Moreover, the customers will lose their confidence toward us. In this study, a harmony search algorithm was presented for this problem is to maximize total revenue. To validate the performance of the proposed approach, we use Harmony Search, Revised Harmony Search, Brute-Force Search to deal with small problem. On the other hand, to deal the big problem, we use HS and RHS to compare their efficiency. The computational results showed that the RHS significantly outperforms the original HS in big problems. It is believed that it can serve as a new benchmark approach for future research.