應用大數據與分群方法發展最佳化儲位規劃與合併揀貨策略-以H公司為例

Applying Big Data and Clustering Methods to Develop an Optimized Strategy for Storage Location Planning and Consolidated Picking -A Case Study of H Company

梁維琦(Wei-Chi Liang)

指導教授 ： 江梓安

國立臺北商業大學/管理學院/企業管理系（所）/碩士(2021年)

摘要

科技的迅速發展使得產品生命週期逐漸縮短。為滿足市場需求之快速變化與強化競爭力，供應鏈必須不斷地提升作業效率。倉儲作業在其中更是關鍵的角色。由於揀貨作業在倉儲活動中最為耗時與費力，管理者無不尋求快速且精確的方法，來提高訂單揀貨效率與降低倉儲營運成本。因此，本研究旨在發展最佳化儲位規劃與合併揀貨策略，以改善揀貨績效。由於儲位規劃是影響揀貨績效的關鍵，故本研究所提出之方法論，結合了儲位規劃與揀貨單分群。首先以顧客訂單預測資料計算零組件需求數量，來預估零組件儲存空間需求，再結合過去大數據資料計算零組件在倉儲平均停留時間，提出最佳化儲位規劃建議。同時，以此儲位規劃為基礎，分別採用階層式分群與K-means分群法，來將相似度高的訂單進行合併。最後，以H公司為案例驗證本論文所提出的方法論。在本研究結果中顯示，優化儲位規劃可以顯著地改善揀貨員檢索零組件在貨架位置的時間。再者，藉由將相似度高的揀貨單合併，可以減少來回揀貨的次數與距離，以達到提升揀貨效率之目標。本研究所提出的改善方法不僅限於電子零組件，此方法之概念亦可應用於其他物料或產品的揀貨作業中。

關鍵字

大數據 ； 階層式分群 ； K-means ； 儲位規劃 ； 訂單揀貨

並列摘要

Rapid technological evolution has led to a decrease in product life cycles, and the efficiency of supply chains must be continually improved in satisfying the rapidly changing market demand and reinforcing company competitiveness. Warehouse management plays a key role in this process. Because order picking is a particularly time-consuming and labor-intensive warehouse operations, managers have sought rapid and accurate methods to improve order picking efficiency and to decrease warehouse operation cost. The goal of this study is to develop an optimized strategy for storage location planning and consolidated picking to improve the order-picking performance. Because storage location planning critically affects order picking, the proposed methodology combines the storage location planning and picking-orders clustering. First, the number of components required is calculated using the customer order prediction data to determine the space required to store these components. Big data are then employed to calculate the average storage time of the components and then to propose the suggestions for the optimized storage locations. Based on the optimized storage location planning, hierarchical clustering and K-means clustering are separately employed to consolidate picking orders with high similarity. Finally, a case study is conducted on Company H to verify the feasibility of the methodology proposed in this study. The results revealed that the optimized storage location planning effectively shortened the picking times for pickers in searching the storage locations of components. Moreover, consolidating the picking orders with high similarity reduced the frequency and travel distances in order picking so as to boost order picking efficiency. The methodology proposed in this study is applicable to not only the order picking for electronic components, but also that for other components or products.

並列關鍵字

Big Data ； Hierarchical Clustering ； K-means ； Storage Location Planning ； Order picking

參考文獻

參考文獻

簡禛富、許嘉裕（2018）。大數據分析與資料挖礦。新北：前程文化事業股份有限公司。

Arai, K., & Barakbah, A.R. (2007). Hierarchical K-means: An algorithm for centroids initialization for K-means. Reports of the Faculty of Science and Engineering, 36, 25-31.

Aristidis, L., Nikos, V., Jakob J. V. (2003). The global k-means clustering algorithm. Pattern Recognition, 36(2), 451-461.

Bahrami, B., Piri, H., & Aghezza, E. (2019). Class-based Storage Location Assignment: An Overview of the Literature. International Conference on Informatics in Control, Automation and Robotics, 1, 390-397.