應用模擬退火演算法求解貨物配櫃利用率之研究

A Simulated Annealing Algorithm Approach to Container Stuffing Problem

非整櫃貨物之併櫃配裝（簡稱配櫃）在目前貨櫃運輸流程中，係一項非常重要且複雜而需仰賴實務經驗的工作，由於貨物的種類與尺寸繁多，配櫃人員往往得耗費許多時間藉助個人經驗或試誤法則來進行排貨和裝櫃，相當不符合經濟效益。因此，在兼顧最佳空間利用率及平穩度下，本研究運用模擬退火演算法之運算架構及特性，加入實際配櫃時所需考量貨物堆疊限制、擺放型態、貨櫃限重與平衡等重要因素，結合改良式「下後左角」優先之貨物堆疊準則，透過MATLAB(R)（數值計算工具加以程式化，並經由參數分析和探討，建構出可用於求解實際貨物配櫃問題的數值模型。同時，亦發展一3D圖像化展示介面，以便能具體展現各貨物之完整堆疊情形。最後經由實例運算結果顯示，於貨櫃空間利用率及整體貨物裝載之平穩度，皆符合配櫃作業所需；且對於多尺寸、不同堆疊限制、組合貨物含棧板，或非矩形體等貨物部分之模擬分析，亦均與現場配櫃情形相當一致，充分展現本數值分析模型在協助實際配櫃作業上良好的適用性及處理能力。

Container stuffing work is a quite important and complicated assignment in container transportation. It is still he accomplished by the workers with actual experiences, but they usually depend mainly on the trial-and-error experiences to load and discharge again and again in order to multiple types and sizes of the cargo could be packed successfully. It may consume much of time and make cost increased. Based on the optimal space utilization and stability conditions, therefore, the purpose of this study is to develop a simulated annealing algorithm approach considering the effects of cargo allocated constraints, container loading limits and stability on packing and integrating with a improved bottom-back-left packing model to solve for actual container stuffing problems. The numerical approach is programmed by MATLAB(R)and a 3D graphic demonstrated interface is also constructed in this study to show the analyzed results specifically. Finally, the computational results indicated that a good container space utilization and stability is good enough to satisfy the requirement of container stuffing assignment. The analyzed results for packing simulation on objects of multiple sizes, irregular shapes, various loading constraints and combination with pallet are agreed well with the practical stuffing work to certify this approach is to be provided with good applicability.