A multiple criteria decision support model based on AHP and GA

碩士論文[[abstract]]從學界關於多準則決策方面的研究文獻中，可以發現一些屬於「決策者的目標函數未知」和「決策者的偏好隨著決策分析進行而改變」的「目標函數未知的多準則決策問題」，還有進一步發展的空間。真實世界中不乏「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「複雜」三者同時存在的多準則決策問題。如何解決這類「目標函數未知的複雜多準則決策問題」是一個相當有趣的議題。本研究將提出一個「以AHP 及 GA 為基礎之多準則決策支援模型」，期盼透過AHP挖掘決策者對於決策準則的偏好，並藉由GA去搜尋可行的方案，能夠解決「決策者的目標函數未知」和「決策者的偏好隨著決策分析進行而改變」的「目標函數未知的多準則決策問題」。本研究所提之模型採多程序模型(multi procedure model)，可支援「邊尋邊選」的策略。換言之，若未找到滿意解，將回復至模型運作之初，反覆運作整個模型，直到決策者找到滿意解為止。也就是說本研究之模型具有學習偏好的能力。本研究的求解過程可分成兩個部分，第一部份：透過AHP求得屬性水準權重；第二部份：將屬性水準權重代入GA中，求出滿意解。由於本研究所提之模型為新創模型，因此本研究旨在進行探討性研究。為驗證本研究所提模型的可行性，我們將以多目標旅遊行程規劃問題為例，進行初步的實驗分析。[[abstract]]Two problems in multiple criteria decision making (MCDM) of objective function unknown, the decision maker value function is usually not known, and decision maker preferences change as the process of analyzing a decision problem progresses, are still to develop further from the academic circle about the MCDM researches. In real world , there is no lack of some MCDM problems about the decision maker value function is usually not known, decision maker preferences change as the process of analyzing a decision problem progresses, and complexity existed simultaneously. How to solve such kind of problem is interesting. This program hopes to solve preceding problems depending on finding out the decision maker''s preference through AHP (analytic hierarchy process), and then searching alternatives by GA (genetic algorithm). This program will offer a multi procedure model, multiple criteria decision support model based on AHP and GA, supporting decision making during search. There are two principal steps about this model, first is to calculate each attribute level through AHP, and second is to find out satisfactory alternatives by GA. At last, we will offer a case about a multiple objective traveling trips problem to test and check preceding model