| 摘要: | 近年來，隨著工業4.0的崛起，數位技術、自動化機械上下料、智慧生產以及物聯網（IoT）的應用等各個方面都經歷了革命性的發展。這些進步不僅在不同產業中帶來了跳躍性的成長，並且有效提高了生產效率並降低了成本。工業4.0的主要目標是將製造資源、數據和流程整合在一起，建立更靈活和智慧的製造產業。在這個過程中，產品裝箱和出貨這一關鍵流程變得至關重要。若能在貨物進行裝箱時有較佳的事前安排，必能使裝載空間達到高效率的使用，如此一來，不僅能節省承載貨物的工具數量（如棧板、貨櫃及貨車等）亦能減少重複嘗試擺放時所費的時間及人力成本。
本文將影像處理的技術用於辨識箱子尺寸，以便將不同種類的箱子進行自動化歸類。並針對不同尺寸和數量的箱子，運用傳統試誤法、貪婪演算法和深度優先搜尋法進行了裝箱運算。透過比較運算結果的差異，並分析三種算法在實際應用中的優勢和劣勢，最終選擇啟發式貪婪演算法作為後續裝箱限制條件實驗應用的選擇。為能更貼近實務需求，本研究提出將裝箱運算中加入重量限制、底面限制和位置限制的條件設定。實驗將機械零件從產出裝入箱子，再堆放至棧板，最終將棧板和貨物一同裝載至貨櫃中，進行一系列的運算和安排。考慮到不同的安全需求，本實驗分別進行了不同限制條件的單獨和組合運算，而加入的限制條件越多，空間利用效率降低的幅度亦越大。
本研究利用Python語言撰寫程式，實驗中利用Matplotlib的工具箱mplot3d將完成裝箱運算的結果繪製成可視化立體圖以便觀察，並讓程式在裝箱運算的過程中同時將執行結果寫入CSV檔，運算結束後可於Excel開檔檢視裝箱成果數據。根據本研究所採用之啟發式貪婪演算法與傳統試誤法求解三維裝箱問題之結果相比較，至多節省約16.20%的可利用空間。
;In recent years, with the rise of Industry 4.0, there has been a revolutionary development in various aspects such as digital technology, automated machinery for loading and unloading, smart manufacturing, and the application of the Internet of Things (IoT). These advancements have not only led to significant growth in different industries but have also effectively improved production efficiency and reduced costs. The primary goal of Industry 4.0 is to integrate manufacturing resources, data, and processes to establish a more flexible and intelligent manufacturing industry. In this process, the critical process of product packaging and shipping has become crucial. A well-optimized arrangement during the packing of goods can lead to highly efficient utilization of loading space. This not only conserves the quantity of tools for carrying goods, such as pallets, containers, and trucks, but also reduces the time and labor costs associated with repetitive placement attempts.
This paper applies image processing technology to identify box sizes for the purpose of automated categorization of various types of boxes. It conducts packing computations for boxes of different sizes and quantities using traditional trial-and-error method, greedy algorithm, and depth-first search algorithm. By comparing the differences in computation results and analyzing the strengths and weaknesses of these three methods in practical applications, the heuristic greedy algorithm is ultimately chosen for subsequent experimental applications involving packing constraint conditions. To better align with practical requirements, this study introduces conditions involving weight limits, base area limits, and position constraints in the packing calculation. The experiments involve loading mechanical parts from production into boxes, stacking them onto pallets, and finally loading the pallets and goods into containers. Considering various safety requirements, the experiments are conducted with different individual and combined constraint conditions. The more constraint conditions are added, the more the efficiency of space utilization is reduced.
In this study, we utilized the Python programming language to write code. During the experiments, we employed the Matplotlib toolbox, specifically mplot3d, to visualize the results of the packing computations in three-dimensional graphs for observation purposes. Additionally, the program simultaneously recorded the execution results into a CSV file during the packing computation process. After the computation concluded, the results can be viewed by opening the CSV file in Excel to examine the packing outcome data. Comparing the results obtained by the heuristic greedy algorithm employed in this study with those obtained by the traditional trial-and-error method for solving the three-dimensional bin packing problem, it was found that the heuristic approach saved up to approximately 16.20% of available space. |
