智慧型多準則決策支援研究：以交談式遺傳演算法為基礎的模型; A Study on Intelligent Multi Criteria Decision Support: IGA-based Model

NCUIR > School of Management at National Central University > Graduate Institute of Information Management > Electronic Thesis & Dissertation > Item 987654321/12814

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/12814

Title:	智慧型多準則決策支援研究：以交談式遺傳演算法為基礎的模型;A Study on Intelligent Multi Criteria Decision Support: IGA-based Model
Authors:	許芳誠;Fang-Cheng Hsu
Contributors:	資訊管理研究所
Keywords:	多準則決策支援;交談式遺傳演算法;決策支援系統;multi-criteria decision support;interactive genetic algorithms;decision support systems
Date:	2000-11-17
Issue Date:	2009-09-22 15:14:45 (UTC+8)
Publisher:	國立中央大學圖書館
Abstract:	由於對決策本質的新體認，多準則決策(Multiple Criteria Decision Making; MCDM)方法也因此興起。從學界關於多準則決策方面的研究文獻中，可以看到多準則決策方法論還存在一些挑戰，特別是面對一些屬於「決策者的目標函數未知」、「決策者的偏好隨著決策分析進行而改變」和「解答空間極大」同時存在的「目標函數未知的複雜多準則決策問題」時，我們發現多準則決策方法應該還有進一步發展的空間。交談式遺傳演算法(Interactive Genetic Algorithms; IGA)除了在本質上適合作為多準則決策模式的核心技術外，還具備其他多準則決策技術少有的特性。本研究從支援決策者進行「邊尋邊選」的觀點，提出一個「以交談式遺傳演算法為基礎的智慧型多準則決策支援模型」解決上述的多準則決策問題。此外，我們發現IGA要能實際運用於支援解決「目標函數未知的複雜多準則決策問題」，必須先克服IGA演化耗時的問題，因此我們進一步提出改善IGA的適應值給定策略。本研究的個案實驗結果顯示：本研究所提模型用於解決二個「目標函數未知的複雜多準則決策問題」研究個案上，其績效顯著的優於傳統What-if模型及IPM (Interactive Programming Method)模型。在IGA演化績效的改善問題方面，個案研究結果也支持本研究所提的分頁「相對距離」策略顯著的優於傳統「全體評比」策略及「偏差」策略。除了上述的貢獻外，本研究模型所提出的隱性基因概念對IGA的應用也深具啟發意義。 Owing to newly recognizing the nature of decision-making, methodologies of multiple criteria decision-making (MCDM) are emerging. After a literature review, it finds that researchers still meet with challenges on MCDM especially for complex multiple decision-making problems with unknown objective function. In this study, we propose a model with interactive genetic algorithms (IGA) to solve the problem. However, the inefficiency problem of IGA needs to be improved to make it feasible for the MCDM problem. Hence, we develop a fitness assignment strategy to improve the performance of the IGA-based system, and integrate them into our model. To verify the outstanding performance of the proposed model, we apply our model on an itinerary planning case. Experiment results show that the model perform as we expected.
Appears in Collections:	[Graduate Institute of Information Management] Electronic Thesis & Dissertation

Files in This Item:

File	Size	Format

社群 sharing

Loading...