Please use this identifier to cite or link to this item: `http://hdl.handle.net/11455/4943`
 標題: 禁忌搜尋法與遺傳演算法混合模式在地下水復育優選問題之應用The Applications of Tabu Search - Genetic Algorithms Hybrid Model on the Optimization of Groundwater Remediation Problems 作者: 林師檀Lin, Shin-Tien 關鍵字: tabu search禁忌搜尋法genetic algorithmshybrid modelgroundwater optimization problem遺傳演算法混合模式地下水復育優選問題 出版社: 環境工程學系 摘要: 地下水復育優選問題為複雜、困難之最佳化問題，以往大多利用傳統之優選技術如非線性規劃法來求解，然此類問題之目標函數及限制式具有非線性及非凸之特性，傳統方法在求解過程中常因其特性，而造成搜尋陷入局部解中難以跳脫。 近十多年來發展出具有強健之尋優能力的啟發式優選技術如遺傳演算法、禁忌搜尋法、模擬退火法等，強調具有跳脫局部解及接受劣化解（inferior solution）之優點，本研究係以具有記憶功能之禁忌搜尋法來求解地下水復育優選問題；並再發展以遺傳演算法與禁忌搜尋法結合之混合模式（GA-TS法）來解決地下水復育系統之優選問題，混合模式擷取了兩個演算法之優點，先以遺傳演算法作前處理器，對問題之解空間作整體性之初步搜尋，再藉由後處理器--禁忌搜尋法之精煉的解題技巧來獲得更高品質的解。 本研究以一虛擬、拘限、均質之地下水復育案例為例，藉由禁忌搜尋法、混合模式針對不同的復育方案予以求解，以了解各優選技術之求解效率，進而提出最具效率之優選技術，供日後地下水之研究者選擇處理技術之參考。 研究結果發現，兼具遺傳演算法與禁忌搜尋法之優點的混合模式，無論是求解何種復育方案，在解的品質與運算時間上，其結果明顯地都較禁忌搜尋法較佳。The optimization problems of groundwater remediation system have the feature that the objective function and constraints may be noncovex as well as nonlinear. Classical gradient-based algorithms including nonlinear programming (NLP) and descent methods have not been successful in solving such kind of problems. In general, these methods converge to a local optimum easily. Heuristic algorithms have becmoe the most widely used tool for solving many optimization problems. These methods, e.g. genetic algorithms (GA), simulated annealing (SA) and tabu search (TS), emphasize powerful and robust optimization procedures. This research selects tabu search for solving groundwater remediation optimization problems. A new hybrid model which combines GA and TS is also developed. In the hybrid model, GA is served as a coarse-grained optimizer, then TS served as a fine-grained optimizer. The results show that the performance of hybrid model is more efficient and robust than TS. URI: http://hdl.handle.net/11455/4943 Appears in Collections: 環境工程學系所

###### 文件中的檔案：

TAIR Related Article

Citations: