多場站車輛路線問題(Multi-depot Vehicle Routing Problem, MDVRP)屬於NP-hard的問題，近年來文獻上通常使用啟發式方法或巨集啟發式方法來求解，以期能在有效時間內得到近似最佳的可行解。本研究以門檻接受法(Threshold Accepting, TA)為基礎，提出八種兩階段的起始解構建方法，並設計可行解調整模組與鄰域搜尋模組，發展出一套可求解MDVRP的巨集啟發式解題架構，TA_MDVRP。為驗證TA_MDVRP之解題績效，本研究蒐集了21題標竿例題，並撰寫電腦程 The Multi-Depot Vehicle Routing Problem (MDVRP) is an extent of the classical Vehicle Routing Problem. More complicated than VRP, MDVRP considers that the usage of multiple depots with fixed size of fleet, and the limitation of maximal duration for each r
