貨就人式揀貨中心的貨架裝填問題

Abstract

本研究針對貨就人式揀貨中心定義了一個考慮數量的貨架裝填問題，當中考慮了每種SKU的庫存量、訂單SKU的訂購量、SKU的體積以及貨架上的SKU裝填數、總容積限制等。此問題源於揀貨中心內複雜的訂單揀貨優化問題。研究目的是針對揀貨中心內的SKU配置，得到最佳的貨架裝填結果，即分配揀貨中心內的SKU至貨架上，以最小化訂單揀貨完成所需的總貨架出場次數。為了得到最佳的貨架裝填結果，本研究研擬了三種啟發式求解法和一種遺傳演算求解法。三種啟發式求解法分別是「熱門度依序裝填法」、「同種類裝填法」和「同種類選取機率依序裝填法」。並實作了一個問題產生及求解系統，該系統包含多種演算程序，可以生成不同複雜度的SKU配置和訂單集合資料，並且可以評估各求解法的貨架裝填結果。通過設計不同規模和情境的標竿問題，我們對以上四種求解法及現今揀貨中心常用的隨機裝填法進行求解比較。測試結果顯示，本研究研擬的求解法在各方面的表現均優於隨機裝填法。此外，三種啟發式求解法和遺傳演算求解法在不同情境下各有優勢。

This work addresses a rack assignment problem considering the quantity of SKUs in parts-to-picker fulfillment centers. It takes into account factors such as the inventory level of each SKU, the order quantities, the volume of SKUs, and the constraints on SKU placements and total volume on the racks. This problem stems from the complex order picking optimization issues within fulfillment centers. The objective is to develop the optimal rack assignment method for configuring SKUs in fulfillment centers to minimize the total number of rack movements required to complete order picking. To achieve this, we propose three heuristic methods and one genetic algorithm method. The three heuristic methods are the " Popularity Sequential Rack Assignment Method," the "Same-Category Assignment Method," and the " Same-Category Selection Probability Sequential Rack Assignment Method." We implemented a problem generation and solving system, which includes various algorithms capable of generating SKU configurations and order sets of different complexities, and evaluating the rack assignment results of each solution method. By designing benchmark problems of different scales and scenarios, we compared the performance of the four proposed solution methods with the commonly used random assignment method in current fulfillment centers. The test results indicate that the proposed solution methods outperform the random assignment method in all aspects. Additionally, the three heuristic methods and the genetic algorithm show advantages in different scenarios.