- 學位論文
考量不確定性的群眾外包配送作業
Last-mile Delivery with Crowdsources Integration Considering the Crowdsources Uncertainty
摘要
摘要 根據研究，群眾外包在終端客戶配送或者稱最後一哩配送(Last-Mile Delivery, LMD)中扮演著重要的角色。與傳統LMD相比，群眾外包運送的主要優點在於其擁有較低的營運成本、資本投資以及運送作業的彈性。就運送成本、服務水準與環境影響而言，將群眾外包運送整合至LMD中已有實質的效益。本研究將群眾外包整合作為LMD運送計劃其中一項的可行的選項。群眾外包提供從轉運點至客戶所在地之運送協助，此整合模式需在轉運點進行主要運送車隊與群眾外包配送者間的包裹轉運。在實際運作情形中，包裹轉運的過程可能會受到各種不確定性因素的干擾(例如：壅塞、天氣變化等)。 本研究將此決策問題區分為兩個角度：確定性與隨機性觀點。就確定性的角度而言，假設在最理想(即沒有不確定性)的情況下，透過將問題規劃成混和整數線性規劃模式(MILP)，藉以研究群眾外包運送的優點。就隨機性觀點來考量不確定性，本研究將群眾外包包裹轉運的成功與否視為導致不確定性的事件，並使用兩階段隨機規劃(SP)來建構最佳化模型以考量此不確定性。除此之外，為處理確定性與隨機性之大規模問題，本研究以禁忌搜尋法(TS)為基礎設計了啟發式演算法。 總體而言，群眾外包運送的整合方式可透過適當分配運送訂單以降低總運送成本，進而改善LMD計劃。然而必須小心達成運送車隊與群眾外包服務之間的平衡，才能達到群眾外包運送合作模式的最大效益。即使將不確定性納入考量，本研究所發展模式依然能達到群眾外包的效益。而透過數值實驗發現，啟發式解法能在快速的運算時間提供高品質的解決方案。 關鍵字：眾包交付、最後一英里交付、兩級路由問題、隨機路由問題
並列摘要
Crowdsource delivery is reported to contribute a significant role for last-mile delivery (LMD). Lower operational cost and capital investment, as well as delivery flexibility, are the main advantages of crowdsource delivery when compared to the conventional LMD. Positive results of integrating crowdsource delivery into the LMD have been reported in terms of delivery cost, service level, and environmental impact. This study investigates the delivery plan of LMD in a collaboration with the crowdsources as one of the delivery options. The crowdsources provide delivery assistance from transfer points to the customer locations. This collaboration requires parcel relay between main delivery trucks and crowdsources at transfer points. In the real situation, this parcel relay activity might be subjected to several kinds of uncertainties (e.g. congestion, weather condition, etc.) that can create disturbance to the process. In this study, the decision problem is tackled from two aspects, the deterministic and stochastic points of view. In the deterministic point of view, the benefits of crowdsources delivery collaboration are investigated given the perfect situation (with no uncertainty) by formulating a problem as a mixed integer linear program (MILP). Upon the uncertainty considered in the stochastic point of view, this study models the parcel transfer or relay event as an uncertain event, which involves the success or failure of the crowdsources’ show-up. A two-stage stochastic MILP model is formulated to as the optimization model considering the associated uncertainty. The heuristics algorithms based on Tabu Search (TS) are designed to handle the large-scale problems for both the deterministic and stochastic versions of the mathematical programming models. In summary, the crowdsource delivery collaboration improves the LMD plan by properly outsourcing some delivery orders to reduce the overall delivery costs. The balance between the delivery fleet utilization and the usage of crowdsourcing service must be carefully achieved to provide the maximum benefit of crowdsources delivery collaboration. These benefits can still be preserved even after the consideration of uncertainty. Based on the numerical experiment, the heuristics algorithm is able to provide the high quality solution with fast computation time.