多重運輸方法最佳化模型

隨著科技日新月異，科技的便利和普及化，使電子商務的應用與發展逐年成長，而近年共享經濟的熱潮席捲，帶動了外送經濟的興起，更將電子商務再次帶起一波高峰，回顧電子商務的發展，為了滿足消費者對購物的即時需求性，建立快速回應且穩定的物流網路一直是電子商務競爭策略下的關鍵，這種物流網絡不僅可以滿足快速增長的共享經濟，還可以滿足電子商務的需求趨勢。隨著國際上外送服務的發展，台灣逐漸發展了更多的即時運送服務運送生鮮等易腐產品，例如食品和飲料運送以及餐飲服務，但與傳統的運輸服務相比，新興的外送服務較不具集中性，舉例Uber Eats或Food Panda等公司僅擁有並運行實時信息系統，而車隊是由個人駕駛員所擁有，因此，他們可以更高的費用提供更靈活，更快速的交付服務。反觀傳統的運輸服務以專業的貨運車隊為中心，並擅長以較低的費用提供大容量，但是，集中式系統在取/送時間表上並不靈活，即使運輸距離很短，交貨也需要一個基本且固定的時間段。在此研究中考慮了台灣電子商務的物流網絡，其中每個弧都提供多種送貨服務，從節點A到節點B的每項不同服務都被視為具有不同特性的不同管道，在面對一批訂單的情況下，目標是以最少總運輸成本，從而獲得必須符合訂購的截止日期的最佳解決方案（運輸方法）。本研究應用作業研究中最小成本流量問題，列出具有特定原節點、接收節點和具有截止時間的多筆訂單，每條管道都具有兩種類型的成本，運費及耗時，並發展出多重運輸管道的數學模型。在少數情況下，此模型可由EXCEL求解，但面臨到中大型問題時電腦求解可能無法運作，因此面對無法估計的大尺寸問題，在實際應用中建議導入啟發式方法。

關鍵字

運輸網路問題；最小成本流量問題；啟發式解法；截止日期；多重圖形

並列摘要

With the rapid development of portable device, it brings convenience and popularization applications and makes the year by year growths of e-commerce. As a result, the boom of the sharing economy in recent years has spurred the rise of the delivery economy and brought e-commerce to another wave of peaks. Beyond the fast growing, the establishment of a quick-response and stable logistics network is the critical key for the future of e-commerce. This type of logistics network meets not only the rapid growing sharing economy but also the demand trend of e-commerce. After the development of international delivery services, more instant delivery services has been gradually developed in Taiwan in many perishable commodities, such as food and beverage delivery and meal services. Comparing to the traditional transportation services, the modern delivery services are decentralized. The companies, such as Uber Eats or Food Panda, only own and run the real-time information systems. The fleets of vehicles are owned by individual drivers. Thus, they can provide more flexible and faster delivery services at higher fee. On the other side, the traditional transportation services are centralized with professional cargo fleets. They are good at providing large capacity at lower fee. However, their centralized system is not flexible on the pickup/delivery schedules. A basic and fixed time period is required for delivery even the shipping distance is very short. In this study, we consider a logistics network of e-commerce in Taiwan, where each arc has multiple delivery services available. Each different service from node A to node B is treated as different pipelines with different characteristics from node A to node B. Considering a batch of delivery orders, the objective of this paper is to minimize the total shipping costs, where the optimal solution (shipping method) must meet the due dates of order. In this study, we adopt the famous minimum cost flow network problem in Operations Research. Each order with specific source node, sink node, and due date are listed. Two types of costs on shipping fee and time consuming are attached to each pipeline and a mathematical model with multiple pipelines was developed. In small cases, the model is solved directed by EXCEL. However, the computer solvers may not be feasible in medium and large cases. For arbitrary large cases, Heuristic approach is introduced for practical applications.