| 摘要: | 由於台灣處於西太平洋颱風侵襲的主要路徑，颱風的侵襲經常帶來強風與豪雨，因此易引發嚴重災情，造成人民生命財產的重大損失。根據消防署統計，近二十四年來平均每年就會發生約7.13次水災。為了防止因水患造成河川兩岸潰堤而影響附近居民的安全，「河川管理辦法」規定河川管理機關應沿河川兩岸之適當地點設置防汛搶險器材儲藏所，作為緊急搶險之用。「搶險」之定義是指天然災害致使河防建造物已發生險象或發生損壞，為防止損壞險象擴大所作之緊急搶救措施。目前防汛搶險器材之佈署常以人工經驗為主，但卻缺乏系統最佳化分析。因此，本研究考量隨機需求下，以最短時間內找出最有效率之調派及成本最小的目標，建構防汛搶險器材佈署模式，期能提供決策者佈署防汛搶險器材於適當的儲藏所之計畫。
本研究藉由時空網路流動技巧建立隨機性需求防汛搶險器材佈署模式，且將隨機性需求修改為平均需求，以建立確定性需求防汛搶險器材佈署模式，此兩模式屬於NP-hard問題。於求解方法方面，確定性模式以數學規劃軟體CPLEX配合C++程式語言進行模式求解；隨機性模式因問題規模龐大，無法直接以數學規劃軟體求解，因此本研究發展一啟發式演算法有效地進行問題求解，並利用隨機相關理論之評估指標以評估模式與演算法之應用績效。最後，本研究參考國內某河川局的實際搶險資料及合理假設輸入資料，進行範例測試且針對不同參數進行敏感度分析。測試結果顯示使用隨機性模式佈署之效益明顯優於人工經驗之佈署方式，故本研究之模式與求解方法可提供河川局相關決策者作為佈署規劃之參考。
;Due to the fact that Taiwan is located right on the pathway of typhoons from the western Pacific Ocean, typhoons frequently bring strong winds and heavy rain and it is easy to cause severe calamities. It might lead to loss and damage of human lives and properties. According to the statistics by the National Fire Agency, on average, there were 7.13 floods each year in the past 24 years. To prevent the levees from bursting and that people who live nearby will in danger, the Regulations on River Management stipulate that, in order to meet the needs in a flooding emergency, river management agencies must set up warehouses on appropriate sites along the river to store flood control and flooding emergency materials. “Flooding emergency” refers to measures taken to stop the situation from worsening as soon as flood control facilities have been damaged. In practice, the decision maker is used to deploy the flood control and flooding emergency materials based on his/her experience, which lakes optimal systematic analysis, Therefore, this research considers the stochastic demand occurring in actual situations, with the aim of optimizing the routing within the shortest period of time and minimizing costs, to construct flood control and flooding emergency materials deployment models. With these models, the decision maker can effectively deploy the flood control and flooding emergency materials at the warehouses.
In this research, the time-space network flow technique is used to construct the stochastic demand and deployment models. We further consider the average demand to construct the deterministic demand model. Both models are formulated as mixed integer multiple-commodity network flow problems, which are characterized as NP-hard. We utilize C++ computer language, coupled with the CPLEX mathematics programming solver, to solve the deterministic model. For the stochastic model, since their problem sizes are too huge to be directly solved by using mathematical programming software. Therefore, we developed a solution algorithm to efficiently solve the stochastic model. We also utilized EVPI and VSS to evaluate the performance of the stochastic model. Finally, we performed a case study using the data collected from a river management office. The test results show that the effectiveness achieved by applying the stochastic model is better than that by pragmatic decisions. The proposed model and solution algorithm could be useful for deploying the flood control and flooding emergency materials. |
