在以顧客需求為導向的生產環境下,產品的設計變的多樣而且複雜,企業為了提高競爭力,整合多個廠區的資源來完成產品的生產,即以跨廠區的環境來規劃產品的生產。在跨廠區的製造環境之下,產品的零組裝次序規劃問題,除了需要考量零件在組裝次序上的限制,零件的組裝工廠也是需要規劃的重點。因此,單廠區的組裝次序規劃已不適用,如何有效率地規劃跨廠區的零件組裝次序以滿足需求,便為研究的重要課題。 本研究提出兩個不同的演算法,以最低組裝成本為目標,規劃跨廠區零件組裝次序。有別於以往將零件組裝次序和組裝工廠指派分為兩部分的求解模式,提出能同時求解零件組裝次序和組裝工廠指派的方法。基因演算法的部份,透過編碼的設計,在基因編碼中混合了代表零件以及工廠的基因碼,使染色體能夠同時表達零件的組裝次序和組裝工廠的指派。粒子群演算法的部份,利用零件組裝權重和工廠指派權重的設計,使粒子位置能夠表達組裝次序和工廠指派。最後,透過實例的驗證,比較兩者在不同特性的問題下的求解效率與求解品質,發現粒子群演算法在搬運成本差異大的情況下能搜尋到較佳的解。
Due to the product design have becoming variety and complexity. The enterprises in order to enhance the competitiveness use the multi-plant planning. The assembly sequence in multi-plant planning, it not only considers the assembly sequence to each component but also assign the component to a feasible plant. Therefore, the single-plant planning for assembly is no longer applied to the problem. How to solve the assembly sequence problem quickly in multi-plant planning has been a key issue. In this thesis, using Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) solve the assembly sequence problem in multi-plant planning. The objective is to obtain optimum the assembly sequence in multi-plant for the minimum cost. The assembly sequence problem in multi-plant planning combines the sequence problem and the assignment problem. We using GA solve two problems at same time by a special encoding rule. And in PSO, we using different definition at multi-dimension in order to solve assignment and sequence problem at same time. Cases are given to show the effectiveness of two algorithm can find the solutions which close to optimal, and characteristics of the algorithm are discussed.