多產品間歇生產調度的MILP模型

電氣自動化論文編號:ZD903    論文字數:16237,頁數:32

目 錄
摘 要 Ⅰ
1 緒論 1
2 間歇過程過程 2
1.1  間歇生產過程及其特點 2
1.2  間歇生產過程與連續生產過程的區別 4
1.3  間歇廠的分類 5
1.3.1  單產品廠 5
1.3.2  多產品廠 6
1.3.3  多裝置廠 7
1.3.4  多目的廠 7
3  生產計劃和生產調度 8
3.1  關于生產計劃和生產調度 8
3.2  生產計劃/調度問題求解的方法 10
3.3  生產排序 10
3.4  間歇調度發展歷程 11
3.5  分支定界法（BAB） 12
3.6  本文模型的優勢 13
4  建立MILP模型 14
4.1  問題的描述 14
4.2  變量定義 15
4.3  數學表達 15
4.3.1  各設備間產品加工先后順序的約束 15
4.3.2  每項任務加工時間約束 16
4.3.3  保證同設備不同任務加工時間不重疊的約束 16
4.3.4  同批次加工任務的順序約束 17
4.3.5  目標函數 17

5  間歇生產調度在玻璃行業的實例 19
5.1  浙江中力控股集團簡介 19
5.2  中力玻璃生產的主要產品及各自工序 20
4.2.1  主要產品簡介 20
4.2.2  產品工藝流程 20
5.3  中力玻璃MILP生產模型 21
5.3.1  模型的前提 21
5.3.2  模型的描述 22
5.3.3  模型的求解 22
5.4  本章結論 24
6  總結 25
致謝 ......26

摘 要

   間歇生產過程，又稱批量生產過程，是工業生產中廣泛采用的一種生產方式，占用很大的比重，而且與我們的日常生活密切相關。間歇過程由于其表現出良好的生產柔性而被越來越廣泛地應用于各種高價值化學產品的生產過程中。然而，對間歇過程的產品和生產成本的評價，傳統觀點基本都還依賴于短期生產計劃。而傳統的生產模型，由于太過于依賴計劃時間的精確度和事件點，因此將大量的時間都花在了對整數變量和約束方程的求解問題上。
 本文提出了一種新的多產品廠間歇調度問題的連續時間混合整數線性規劃(MILP)模型，該模型的整數變量體系不依賴于時間塊(或者事件點)的概念，并且利用了變量物理概念上的對稱互補性，使得與傳統的建模方法相比不僅整數變量的數目減少了一半以上，而且建模思想、建模理論都有了新的改進。數學規劃是研究間歇生產調度的主要方法之一。本文分析了順序型多目的工廠間歇生產的特點，建立了一種新的混合整數線性規劃 (MILP)模型。通過對一個算例的考察，證實了新模型可以快速地求得全局最優解。并應用了一個實際例子，證明MILP模型在玻璃加工業（間歇生產）計劃調度上的明顯優勢。
 
 
關鍵詞：間歇過程調度  多產品廠  混合整數線性規劃  玻璃加工業 分支定界法
Abstract
 Batch production process, also known as the mass production process, industrial production is widely used in a productive manner, the occupation of great proportion, but also with our daily lives are closely related.Batch processes are widely used in producing many high-value chemical products for its good flexibility. However,the production and cost effectiveness of batch plants mostly depend on short-term scheduling. The construction of traditional models for batch scheduling relies on the definition of time slots or time events．which lead to the result that many binary variables are involved in the models and it takes much time to solve the problems with the branch and bound method.
 A new kind of model for scheduling of multi-product batch plan ts is presented ．The new model uses a continuous-time domain representation that does not rely on the definition of time slots or time events．The symmetrical and complemental feature of binary variables is also considered to simplify the model．As a result, the number of binary variables of the new model is successfully reduced to no more than half of the traditional ones ．A medium-scale example is presented  to prove that the new model quickly solve and reach the global optimal result．Comparisons with traditional model are also provided with large-scale problems．Moreover，the principle of the new model can be not only used in the multi-product problem，but also easily applied in other aspects such as multi-purpose batch scheduling and so on． And application of a practical example,To prove the obvious advantage of MILP model on the planning and scheduling in the glass industry(Batch)

Keywords：batch process, scheduling, multi-product plant, MILP, glass industry, branch and bound method