Please use this identifier to cite or link to this item:
標題: 遺傳演算法於配水管網設計最佳化之應用
The applications of genetic algorithms on the optimal design of water distribution network
作者: 梁家明
Chia-Ming, Liang
關鍵字: Genetic algorithms (GA)
Water distribution network
出版社: 環境工程學系
摘要: 中文摘要 自來水建設工程為都市中必要的公共建設。一般而言,自來水的工程建置可分為集水、輸水、抽水、淨水及配水等項目,其中以配水管網建置費用最高,據陳(1997)指出建置配水工程的費用就佔整體自來水總工程費用之三分之二以上,故若能在配水管網建置前就先妥善規劃,將能節省一筆可觀的建置成本。 配水管網最佳化設計屬大尺度、離散性且複雜化的問題,早期採用傳統技術方法不但耗時且往往只能求得局部最佳解,對於設計規劃而言,不符經濟性。若以近年來發展的啟發式(heuristic)演算法,取代傳統方法求解配水管網的最佳化問題,其結果顯示啟發式演算法除在求解效率提升外,獲得的解也近似全域最佳解。故本研究以啟發式演算法中的遺傳演算法(Genetic Algorithms, GA)為基礎,採取實數編碼(real-number coded)方式且加入新的機制發展兩階段RGA(real coded genetic algorithms)模式與水力分析模式結合,來求解四個配水管網案例之最佳化問題,並與文獻中所提及的測試結果做比較。 本研究分別以兩階段RGA模式及單純RGA模式對配水管網案例進行求解,結果可得兩階段RGA模式的求解能力較單純RGA模式優異,與文獻進行比較的結果亦顯示兩階段RGA模式在求解效率及解的品質等方面,都有不遜於其他優選方法的表現。
Abstract The optimization of water distribution networks are complex problems that can hardly be solved with conventional optimization algorithms. This study had developed a real number coded genetic algorithms (RGA) to solve these problems. Additionally, another so-called “Two phases Real number coded Genetic Algorithms (TRGA)” was also employed in this study. Several well known benchmark optimization problems of water distribution networks are used as case studies to evaluate the performance of these GA models. The optimization results show that the TRGA is able to find solutions comparable to those provided by some of the most competitive algorithms published in the literatures, in terms of both the computational efficiency and solution quality.
Appears in Collections:環境工程學系所



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.