Search Algorithms in the Selection of Warehouses and Transshipment Arrangements for High-End Low-Volume Products

高價低量商品倉儲選擇及轉運調度之搜尋演算法

洪一峯(Yi-Feng Hung) ； 陳威志(Wei-Chih Chen) ； 陳建良(James C. Chen)

《先進工程學刊》 7卷2期 (2012/04) Pp. 51-60

https://doi.org/10.29948/JAE.201204.0001

摘要

在當前競爭激烈及低利潤之市場條件之下，連鎖零售產業所面臨的挑戰是如何管理庫存以降低成本並改善客戶服務，以爭取更大的市場占有率。在本研究提出了一個新的轉運模式，從現有的零售點中選擇部分地點作為高價低量產品之倉庫，並負責這些產品之存貨管理及配送作業，藉由風險分擔的效果，最小化系統整體成本。然而，產品儲存點的選擇必須與庫存管理及轉運安排一併考量以達成系統整體之最佳化。本研究所提出之轉運模式可用一個非線性整數規劃問題表示，並發展了以模擬退火法及塔布搜尋法為基礎的二個區域搜尋法來求解非線性整數規劃問題，以解決傳統數學方法求解時間過長與無法求解大規模問題之困難。研究結果顯示，在小規模問題上模擬退火法與塔布搜尋法均可快速的求得最佳解。本研究將測試問題擴大至200個零售點，以測試大規模問題之下二個方法之求解效能，實驗結果顯示在大規模問題上，塔布搜尋法之求解時間及求解品質表現優於模擬退火。

關鍵字

供應鏈管理 ； 風險共擔 ； 選址－庫存模型 ； 模擬退火法 ； 塔布搜尋法

並列摘要

In this highly competitive and low-profit marketing environment, the challenge that chain-store resale industry face is how to implement inventory management techniques so that cost down is possible and quality of customer service can be improved simultaneously to broaden market share. In this study, a transshipment approach is proposed by selecting several retail stores not only served as a warehouse for high-end, low-volume products but also are responsible for inventory and delivering to share and lower their risk to minimize overall system costs. However, locations of warehouse to products must be considered along with inventory management and transshipment arrangements to achieve their maximum performance for the whole system. To solve this problem, a transshipment model, proposed in this research, is formulated by an integer nonlinear programming methodology. And two efficient local search algorithms based on Simulated Annealing and Tabu search are developed on integer nonlinear programming, instead of using traditional mathematical approaches, having difficulties in solving these problems with time-consuming drawback and restrictions on large scale. Study shows that both algorithms are efficient, obtaining optimal solutions on small-sized problems. In this research, computational results with up to 200 retail stores to evaluate the performance with these two algorithms in large scale are also presented. And based on the acquisition of data, Tabu Search outperforms Simulated Annealing either in time-consuming or calculating quality.

並列關鍵字

supply chain management ； risking-pooling ； location-inventory model ； simulation annealing ； tabu search