| 摘要: | 在台灣保全公司負責銀行各分行現金護運至總行、各提款機現金之護運、有簽訂契約之公司行號(如全聯福利社、家樂福、大潤發…等)現金護運至銀行時，經常面臨被搶劫的風險，每當被歹徒搶劫往往會對保全公司之客戶端造成重大的財物損失。台灣過去十年間，運鈔車被歹徒搶劫之事件就高達 35 件之多，平均每年會有 3.5 輛運鈔車被歹徒搶劫，頻率非常之高，從許多運鈔車被搶劫的案例中可以發現，運鈔車護運作業行駛的路線與抵達時間欠缺變化性是導致被歹徒輕易的跟蹤、觀察，進而掌握行駛的動向的主要因素。目前實務排程方式是經由人工排程的方式規劃出幾條可選方案進行指派，此作法缺乏系統分析，因此規劃結果經常造成資源浪費。此外，人工排程的方式並未考慮到運鈔車護運路線缺乏變化性這一點。緣此，本研究構建一運鈔車護運作業排程模式，以保全公司營運成本最小化為目標，考量運鈔車每日護運路線所需的變化性，構建一時空相似度幫助決策單位之決策者規劃每日護運路線時考慮到先前所行駛過的護運路線與抵達時間，盡可能規劃出與先前護運路線不相似之護運路線，增加運鈔車護運路線與抵達時間的變化性與安全性。 本研究利用時空網路流動技巧建立此模式，以定式車輛在時空中的流動情況，屬 NP-hard 問題。在求解方法上，利用 C++ 程式語言配合數學規劃軟體 CPLEX 進行模式求解，但當面臨實務的大型問題時，勢將難以在有限時間內利用數學規劃軟體求得最佳解。緣此，本研究發展一啟發式演算法以有效地求解問題。最後為評估模式與演算法之實用績效，本研究以國內一保全公司的營運資料以及合理假設產生測試範例，進行範例測試並針對不同參數進行敏感度分析，結果顯示本模式與演算法在實務上可有效的運用並提供保全公司作為參考。 The security companies in Taiwan are mainly in charge of escorting (and delivering) the cash from branches to each major bank and ATM, and also from the companies (i.e. superstore, mall…etc.), which had contract with the security companies, to some specific banks. When the guards from the security company is in charge of the escorting mission of cash-transportation truck, they often faces the risk of robbing, which always causes great property of damage to the clients. In the past 10 years, the robbing incidents in Taiwan have reached up to 35 cases, with an average of 3.5 cars per year being robbed. From the recent cases, we have discovered that, the lack of variability of the route and arriving time of cash-transportation truck is the main factor causing the robberies. Bandit can easily trace, observe, and master the driving tendencies of each truck. Present practical scheduling model is used for artificial scheduling to plan several modes of feasible plans and designs. This mode lacks the systematic analysis which often causes the waste of resources for the resulted scheduling. Besides, the artificial scheduling also lacks the consideration of route variability for the trucks. Therefore, this research constructs a scheduled model for cash-transportation truck, which accounts for the variability of the routes and minimizes the operating costs of security companies as much as possible. Therefore, the research designs parameters of dissimilar space-time, which hoped to help the decision-makers plan the daily route and arriving time as different as possible, and to increase the variability and security of the routes and arriving time for cash-transport truck. We employ network flow technique to develop the model, which is characterized as NP-hard to formulate the vehicles in the dimension of time and space. We use C++ and CPLEX to solve the problem, but it is difficult to solve in a reasonable time when problem size is huge. Therefore, we develop a heuristic for solving problem effectively. Numerical tests based on real operating data from a security company and reasonable example to evaluate the proposed solution algorithm. The preliminary results could be useful for security companies. |
