以粒子群演算法求解具潛在性訂單之供應鏈生產問題

Apply Particle Swarm Optimization Approach to Solve a Supply Chain Production Problem with Potential Order

陳美利(Mei-Li Chen)

指導教授 ： 宮大川

中原大學/電機資訊學院/工業與系統工程研究所/碩士(2009年)

https://doi.org/10.6840/cycu200901610

摘要

摘要
面對競爭愈來愈激烈的市場，企業選址設廠也從以往的區域性轉為全球性，透過全球化之發展以期望服務更多的顧客，且為了求快速提供顧客完善的服務，將其產品採取混合生產模式進行規劃，其標準化程度高、存貨成本較低的產品採取計畫式生產，技術性高、存貨成本較高的產品則採取組裝式生產，透過此一混合生產模式客以快速回應顧客的需求，提供更快速的服務，提昇企業自我競爭力。
本研究探討對象為機械產業，因產品機台單價昂貴，顧客於下單前皆會多方考量各項因素，並與企業洽談多次後才進行確定下單，其在洽談過程中之訂單稱為潛在性訂單，確定下單時間之訂單稱為確定性訂單，然而面對顧客無法確定實際下單時間的情況下，企業也無法快速做出完善的事先預測規劃。因此，本研究提出一潛在性訂單之訂單規劃機制，透過模擬的概念隨機產生顧客訂單之數據，觀察每期顧客訂單成交機率變化，判斷當期各潛在性訂單之機率值是否大於設定之門檻，大於門檻值將給予訂單適當的產能預留，而其產能預留將影響產品生產的順序。於此，本研究也建構了一訂單規劃數學模式，並藉由粒子群演算法搜尋其各訂單配置之適當生產廠區與時間，比較在交期、機率乘上工時貢獻度兩種不同排序策略下之規劃對訂單安排所造成的影響，依範例數據分析所得結果，我們發現在面對訂單排序時考量到工時貢獻度以及成交機率可得到較大的利潤。

關鍵字

粒子群演算法 ； 訂單規劃 ； 潛在性訂單 ； 排序策略

並列摘要

The enterprise selects its plant location from a regional perspective now to a global one when facing the competitive market. By the development of globalization, enterprise also hopes to offer a better service to the increasing customers. To provide the better service, the enterprise adopts the mix production model. When manufacturing the high standard products and ones with low cost in inventory, the MTO production is adopted. The CTO production is used when a product is manufactured with high technique and high inventory cost. The mix production model can serve the customers in a timely manner and promote the enterprise competitiveness as well.
The research object is the mechanical industry. Customer will place an order in every aspect due to the high cost of the machine. The order is thus called a potential order. In addition, an order placed on time will be named as a confirmed order. However, the enterprise is still unable to offer a plan without the customer’s on-time order.
The research proposes a plan that supports the potential order. Stochastic data of the customer’s orders will be generated as well. And the change of the customer’s successful probability will be observed to see if the probability of the potential orders reaches the threshold of the hypothesis. If the data is higher than the threshold value, the order capacity can be reserved and the reserved capacity will also affect the production sequence.
The research also constructs a mathematical model for the scheduled order and the particle swarm optimization will be used to calculate the production time and value the location of the factory to produce in. Two kind of different strategies of the delivery date, probability that contribution of working-hour multiplies by will be compared to see the effect to the scheduled order. According to the data researched, a better profit will be obtained when the contribution of working-hour and successful probability are both considered.

並列關鍵字

Particle swarm optimization ； Schedule strategy ； Orders planning ； Potential order

參考文獻

[10]張益菁，「考量需求不確定之單階多廠產能規劃問題—以TFT-LCD產業為例」，國立清華大學工業工程與工程管理學系碩士論文，2007。

[13]莊富鈞，「以BOM為基之二階供應鏈生產環境分析」，中原大學工業工程學系碩士論文，2003。

[19]詹益勳，「MTS/CTO環境下之多產品多廠區供應鏈生產策略分析模式發展」，中原大學工業與系統工程學系碩士論文，2009。

[21]劉維展，「模組化供應鏈與生產環境策略之分析模式發展」，中原大學工業工程研究所碩士論文，2004。

[3]吳志武，「晶圓代工廠二階段訂單允諾機制」，國立清華大學工業工程與工程管理學系碩士論文，2007。