生產系統之動態分解派工方法

Abstract

本篇論文研究開放式列隊網路 (open Queueing networks)的指派問題。通過利用結合數個動態規劃結果，達到近似最佳生產系統控制模型，用以精進控制系統結果。動態規劃由於受制於維度詛咒及重建模型問題，在求解較大規模系統的最佳控制策略時往往會花費很長時間，而且每當新設置或製程出現就必須重新建構數學模型。然而，動態最佳控制策略存在著一定規律，零工問題本身也存在一些特性。若利用這些性質於方法之中，將小規模系統的最佳控制策略結合，用來預測大規模系統的最佳控制策略，將可以克服因為利用動態規劃求解最佳策略亦或是重新建模等所花費的時間成本。

This study presents a dynamic approach method for Multiple stage job shop manufacturing systems. Due to the computational complexity and memory requirement, dynamic programming cannot efficiently find optimal control policies for realistic operating systems which usually contains large number of machines and productions. It often takes a long time to solve the optimal control strategy of a large-scale system, and the mathematical model must be reconstructed whenever a new setting or process occurs. However, with some properties of the problem we can find an easy way to cope with these problems.