以基因演算法求解汽車零件倉庫揀貨路徑之研究

隨著近年汽車產業的技術進步，汽車使用年限增加，新車銷售額呈現負成長，現今各家廠牌的車廠已不像以往將新車銷售當作主要收入來源，而是將重心放在消費者更為重視的售後維修服務。在售後維修服務的營業收入來源主要分為服務工資及零配件銷售，而根據售後維修服務部的統計汽車零件營收額更是佔了售後維修服務部的50-60%，這凸顯「零件倉庫」是個可以增加營收不可或缺的重要單位之一，除了能即時供應現場保養維修、通告召回和保固零件更換外，更能大幅增加整個車廠的收入來源，因此如何管理好零件倉庫便成為了一個重要的課題。本研究將針對影響揀貨效率中的「揀貨路徑規劃」來為此零件倉庫進行效率提升，還能適用於此研究之P倉庫只能以訂單別揀取的限制，利用基因演算法來解決關於這種需要在離散變數中找尋最佳解的問題，相較其他演算法可以避免陷入局部最佳解的狀況。使用本研究之演算法後，得到的研究結果顯示不僅在揀貨作業時間中縮短15-20%的時間成本，人員能用統一且有效率的方法揀貨時也能降低失誤率，且在針對新人的訓練也能快速的在揀貨這項工作上手。

關鍵字

基因演算法；路徑規劃；汽車零件；訂單揀取

並列摘要

With the technological advancements in the automotive industry in recent years, the lifespan of automobiles has increased, leading to a negative growth in new car sales. Nowadays, automotive manufacturers no longer rely solely on new car sales as their main source of revenue, but instead focus on the aftermarket maintenance services, which consumers value more. The revenue sources of aftermarket maintenance services mainly consist of service labor and parts sales. According to statistics from the aftermarket maintenance service department, automotive parts sales account for 50-60% of the department's revenue. This highlights the importance of the "parts warehouse" as an indispensable unit that can increase revenue. In addition to providing timely supplies for on-site maintenance, recall notices, and warranty parts replacement, it can significantly increase the revenue source of the entire car factory. Therefore, managing the parts warehouse effectively has become an important issue. This study aims to enhance efficiency in this parts warehouse by focusing on "picking path planning" that affects picking efficiency. It addresses the constraints of a P-type warehouse that allows picking only by order types. Using genetic algorithms to solve this problem of finding the optimal solution in discrete variables, the study avoids the pitfalls of local optimal solutions that other algorithms may encounter.Results from the algorithm implementation show a reduction in picking operation time by 15-20%, thereby cutting costs. It also enables personnel to pick items more uniformly and efficiently, reducing error rates. Additionally, it facilitates quicker training of new personnel in the picking process.