混合流程型(Hybrid flow shop; HFS)生產是結合流程式生產排程(Flow shop scheduling; FSS)與平行機台(Parallel machine scheduling; PMS)所構成的生產環境，研究指出此種類的生產環境中，必須在考慮訂單的加工順序下，同時還須考慮其如何分配訂單至機台之派工問題，為NP-complete問題。此外，HFS問題從過去一張訂單僅在單一生產機台加工的情況，到後來的一張訂單能規劃多機台進行生產(Multiprocessor task)，目前的研究多著重在已知各訂單之機台資源配置的情況下，以進行規劃，然而，隨著產業製程的彈性需求增加，目前生產環境已要求在機台資源未事先配置下，進行最佳化的生產排程、機台資源配置與機台派工。因此，本研究將以一張訂單在未知情況能分割並配置於多機台為基礎下，同時在多階段製程加工與多平行機台的生產方式下，透過「整合變動鄰域搜尋法和粒子群最佳化演算法」以最大完工時間最小化(Makespan)為目標，用於求解訂單生產順序與決定每張工單應配置到哪些機台上，並將此研究應用在晶矽太陽能電池製造業上，在考量晶矽太陽能製程特性，包括平行機台加工、專用機台、獨立整備時間與相依整備時間之特性，使生管人員能以Makespan為目標下，求得的訂單生產順序與各訂單之機台資源配置為作最佳排程規劃的依據。最後並建構出相同問題定義的數學模型，與本研究所提出的演算法進行比較與分析，其最後實驗結果顯示，大幅優於其數學模型求解之品質與時間。 The Hybrid flow shop scheduling problem may be seen as a generalization of two particular types of scheduling problems: the ?ow shop scheduling (FSS) problem and the parallel machine scheduling (PMS) problem. The key decision of the ?ow shop scheduling is the sequence of orders through the shop whereas the key decision of the parallel machine scheduling problem is the allocation of orders to machines. This problem is NP-Complete problem. In addition, the HFS problem is one job on one machine pattern at the earliest. HFS problems can be one order processing more than one machine at the afterwards. (Multiprocessor task). Most of current researches focus on orders which are given allocated production resources. However, with the industry's increased demand for process flexibility, the current production environment has been requested resource is not pre-configured machine. Production scheduling, machine resource allocation and machine dispatching are optimized. Therefore, this research will be split into several batches when the unknown with one order, and the order allocated to multiple machines, while processing in multi-stage process with the parallel-machine mode of production. Through the proposed method to minimize the makespan, solving order sequences and each work order can be configured to decide to which machine. This research will be used in solar cell industries. This research will consider the characteristics of the solar cell manufacturing process, including parallel processing machine, machine eligibility, sequence independent setup times and sequence dependent setup. Production management staff can minimize the makespan for the goal. Then find the sequence of orders and all orders of the machine resources as the basis for optimal scheduling. Finally, we will construct a mathematical model of the same problem definition, and compare with our algorithm and analysis of performance. The results showed significantly better than the mathematical model to solve the quality and time.
