Please use this identifier to cite or link to this item:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/88434
Title: | 多產品彈性流線型環境中的預防性保養與生產規劃問題 Preventive Maintenance and Production Planning in a Multi-product Flexible Flow Shop Environment |
Authors: | 張若瑜 Jo-Yu Chang |
Advisor: | 孔令傑 Ling-Chieh Kung |
Keyword: | 彈性流線型排程,預防性保養排程,生產隨機性,數學規劃,啟發性演算法, flexible flow shop,preventive maintenance scheduling,production uncertainty,mathematical programming,heuristic algorithm, |
Publication Year : | 2023 |
Degree: | 碩士 |
Abstract: | 在流線型生產排程問題中,將機台保養決策納入考量是一個相當困難的權衡問題。一方面,長期不進行機台保養會導致機器狀況惡化,增加生產過程中的不確定性,進而降低系統的生產產量和品質。然而,進行維修保養也意味著需要使用佔據機台的工作時間,這可能會延宕既定的生產計畫,並可能導致短期內的需求被犧牲。因此,在多站點且每個站點都有多個平行機台的情況下,如何做出最佳的生產和維護決策以最大化整體效益,是一個值得深入討論的重要問題。
本研究針對多產品彈性流線型環境討論生產與保養排程的聯合調度問題。在我們的問題中,機台的狀態會影響生產數量的隨機性,而進行保養排程可以使機台恢復到較佳的產能。本研究的主要目標是最小化缺貨成本和存貨成本之和。我們使用動態規劃對問題進行建模。由於動態規劃模型無法在實務上的可接受時間內找到最佳解,因此我們提出了兩種啟發式演算法,一種針對機台在單一時間區段內可多工的情境進行討論,另一種針對僅能單工的情境進行調整。演算法可以分為三個階段,包括生產規劃、保養排程和生產規劃的微調。該演算法以滾動視窗的方式在每個時期的開始重複執行。 為了證明這兩種啟發式演算法的有效性和效率,我們設計了九種情境和三個擴大問題規模的面向,並使用我們的演算法進行實驗。實驗結果表明,我們的演算法可以在有效地減少計算時間的同時獲得近似解,並且尤其適用於瓶頸站明顯、產能吃緊且供不應求的情境。最後,我們也將合作企業的實際案例套用至我們的模型中,進一步證明我們的演算法兼具高效及實用性。 Balancing production and maintenance scheduling poses a challenging dilemma. Prolonging maintenance intervals can lead to poor machine conditions, affecting production quality. However, performing maintenance may disrupt the production plan and sacrifice immediate demand. Therefore, finding the right balance and timing between production and maintenance is a crucial and thought-provoking issue. In this study, we address a joint production planning and preventive maintenance scheduling problem in a multi-product flexible flow shop environment. The machine conditions have an impact on production uncertainty, and performing maintenance can improve the machine condition. The objective is to minimize the sum of shortage costs and inventory costs. Since obtaining an optimal solution using a dynamic programming model is time-consuming, we propose heuristic algorithms to solve our problem, one for multi-tasking and the other for uni-tasking scenarios. This algorithm can be divided into three phases, which are initial production planning, maintenance scheduling, and production plan adjustment, and should be repeatedly conducted at the beginning of each period. To prove the effectiveness and efficiency of our algorithms, we design scenarios with four factors and three extension aspects for testing purposes. The results demonstrate that the proposed algorithm can efficiently reduce computation time while obtaining near-optimal solutions, and it performs especially well in scenarios with limited capacity and a clear bottleneck station. Lastly, we display the application of our algorithm in a realistic case. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/88434 |
DOI: | 10.6342/NTU202301901 |
Fulltext Rights: | 同意授權(全球公開) |
Appears in Collections: | 資訊管理學系 |
Files in This Item:
File | Size | Format | |
---|---|---|---|
ntu-111-2.pdf | 2.87 MB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.