透過您的圖書館登入
IP:3.21.244.14
  • 學位論文

人工智慧演算法應用於藥劑噴灑最佳化問題之研究

Artificial Intelligence Approaches for the Optimal Spray of Liquid Medicament

指導教授 : 謝益智
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


近年來登革熱(Dengue)在各地肆虐,其中尤以台灣高屏地區為登革熱之高危險地帶。登革熱屬於地區性傳染疾病,一旦蔓延將造成大規模傳染與民眾恐慌,因此登革熱防制格外重要。本研究探討藥劑噴灑最佳化問題,此問題包含每週可用藥劑、病媒蚊擴散比率與工作團隊數的選擇,問題目的為如何有效地噴灑有限的藥劑以降低傳染病的蔓延,以使疫情控制在安全範圍內。 本研究探討的問題假設各地區病媒蚊指數將依照其臨近接觸面積前一週病媒蚊指數的情況而會不斷地上升,唯有進行藥劑噴灑才可降低病媒蚊指數,因此本研究之問題複雜度相當高。若以傳統演算法來求解此藥劑噴灑最佳化問題,其運算時間將較為冗長,並且無法求得最佳解,因此具有求解快速與能夠得到近似最佳解之人工智慧演算法便成為一個較可行的方式。 本研究以提出之免疫演算法(Immune Algorithm,IA)、基因演算法(Genetic Algorithm,IA)與粒子群演算法(Particle Swarm Optimization,PSO)等三種人工智慧演算法對此藥劑噴灑最佳化問題進行求解。除此之外,為了確保求解品質,本研究採用田口方法中直交表的應用進行實驗設計,以找出最佳參數來求解此藥劑噴灑最佳化問題。本研究分別對以里作為目標的旗津區與以區作為目標的高雄市其中20個區進行測試,數值結果顯示,免疫演算法的求解品質優於基因演算法與粒子群演算法。除此之外,本研究亦探討相關參數對疫情的影響,例如:總藥劑數量、病媒蚊擴散比率與工作團隊數等。

並列摘要


There have been several cases of dengue in Taiwan recently. Kaohsiung and Pingtung are the high risk areas of dengue in Taiwan. The dengue is a kind of regional infectious diseases, and if we can not control this disease, then it will cause serious infection and people’s panic. So the spay of liquid medicament is more and more important. This thesis studies the optimal spray of liquid medicament problem which contains several important issues to be decided, e.g., how to allocate the weekly available liquid medicament and the number of teams for various diffusion rates of mosquitoes. In this thesis, we assume that the index of mosquito of a specific area will continue to rise based upon the indices of mosquito of itself area and of its near regions. The only way to reduce the index of mosquito of a specific area is to spay the liquid medicament. Therefore, the complexity of the problem in this study is quite high. Typical mathematical programming approaches are time expensive for finding the optimal solution of this problem. Therefore, in practice, heuristics are proposed to solve this optimal spray of liquid medicament problem. In this study, we propose three heuristic algorithms to solve the spray of liquid medicament problem, including immune algorithm, genetic algorithm and particle swarm optimization. In addition, in order to ensure that the solution quality, in this study, we use the orthogonal array of Taguchi method to solve for the best parameter of this sprayed optimization problem. In this study, we test two problems in Chichin and Kaohsiung separately. Numerical results show that the solutions by immune algorithm is better than those by genetic algorithm and particle swarm optimization.

參考文獻


4.行政院衛生署疾病管制局,「瘧疾個案殺蟲劑噴灑標準作業流程」,參考日期2011/6/26。
5.李旺蒼,2006,以粒子群最佳化為基礎之混合式全域搜尋演算法求解含凹形節線成本最小成本轉運問題之研究,國立中央大學土木工程學系,碩士論文。
7.林政宏,2007,台灣地區登革熱擴散之空間分析,國立臺灣大學理學院地理環境資源學研究所,碩士論文。
8.林豐澤(2005),「演化式計算下篇: 基因演算法以及三種應用實例」,東北大學資訊科學與工程學院系統工程學報,3卷,2期,頁29-56。
10.夏維泰,2009,建置登革熱病媒蚊化學防治資料庫(第一年)研究報告,行政院衛生署疾病管制局98年度科技研究發展計畫。

被引用紀錄


陳維翰(2016)。應用人工智慧演算法於生產線平衡中的工作站規劃問題〔碩士論文,國立虎尾科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0028-2407201611234900

延伸閱讀