Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/5111
標題: 強化螞蟻演算法與禁忌搜尋法之混合模式於配水管網設計最佳化之應用
The Applications of Enhanced Ant-Tabu On the Optimal Design of Water Network System
作者: 吳宗樺
ZONG-HAU, WU
關鍵字: 配水管網;water distribution;最佳化;螞蟻演算法;啟發式演算法;禁忌搜尋法;optimization;Tabu Search;Ant-Tabu;ant system;heuristic algorithms;water distribution;optimization
出版社: 環境工程學系
摘要: 
配水管網最佳化設計屬大尺度、離散性且複雜化的問題,以往採用傳統性的優選方法不但耗時,且往往求到的解為局部最佳解。因此,有不少學者拿近十幾年來才開始蓬勃發展的啟發式(heuristic)演算法,取代傳統方法求解配水管網的最佳化問題,也獲得了相當不錯的成果。故本研究亦融合了螞蟻演算法與禁忌搜尋法兩種啟發模式,並強化其移步及相關機制,發展為進化螞蟻-禁忌之混合模式(Enhanced Ant-Tabu, EAT),用以求解配水管網最佳化問題,並與文獻結果進行比較。結果顯示,EAT模式在多峰函數以及配水管網問題的求解上皆較現有文獻為佳,除了目標函數運算次數得以降低之外,求解品質亦有所提昇。

Several classical optimization problems of water distribution networks were successfully solved by a so-called “Enhanced Ant-Tabu (EAT) System” developed in this research. EAT is a stochastic meta-heuristic method, which combines ant system (AS) with tabu search (TS) to solve optimization problems. It employs the foraging behavior of ants as well as human memory systems to avoid the risk of trapping into local optima. Besides, this research addresses a new move strategy designed to reinforce the ability of searching global optimum. A series of optimization tasks of multi-modal, multi-dimensional functions were performed to evaluate the global optimization ability of EAT. The results indicate that EAT outperform other optimization algorithms proposed in relevant literatures in terms of both the solution quality and computational efficiency. Furthermore, EAT is also found an outstanding methodology for solving water distribution network optimization problems.
URI: http://hdl.handle.net/11455/5111
Appears in Collections:環境工程學系所

Show full item record
 

Google ScholarTM

Check


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