- 學位論文
結合證據理論的層級分析法於多準則決策之研究
A Study of Multiple Criteria Decision Making for Integrating Dempster-Shafer Evidence theory and AHP
摘要
層級分析法（Analytic Hierarchy Process, AHP）為一種非常普及的多準則決策之方法，主要能處理或衡量問題與準則之間的衝突，並且透過綜合性評估整體問題，提供最佳的問題解決方案與可行方案的優先順序，但AHP往往令人詬病的問題在於無法處理不確定性、具有模糊性問題，以及成對比較次數過多造成計算過程繁雜與問卷題目激增，進而讓使用者思緒混淆。Beynon and Morgan（2000）提出結合Dempster-Shafer證據理論與AHP的方法（DS/AHP），對AHP成對比較次數進行改善，結果顯示可大幅減少比較次數，但是Dempster-Shafer證據理論當衝突係數過高會影響結果的合理性與一致性之問題，同時可能DS/AHP演算法之內，可能會影響整體所產生之方案推薦結果，造成不合理狀況。 本研究利用新型態Dempster-Shafer證據理論合併規則，針對傳統的DS/AHP內衝突係數過高之問題進行改良。本研究並開發雛形系統，用以評估演算法的績效，並與AHP與傳統DS/AHP進行分析比較。結果顯示本研究改良後的DS/AHP能有效降低衝突係數與整體計算次數，減少發生不合理的推薦結果。
並列摘要
AHP (Analytic Hierarchy Process) is a very popular method for multiple criteria decision making problem. It can deal with the conflict between the criteria, comprehensive assessment of the decision problem, and provide a suggestion to decision maker. However, AHP is often criticized for dealing with uncertainty or ambiguity problem, as well as the huge number of pairwise criteria comparison. Beynon and Morgan (2000) combined Dempster-Shafer evidence theory and AHP, and the new method is named as DS/AHP. DS/AHP can reduce the number of criteria comparison. However, in the Dempster-Shafer evidence theory, the high conflict coefficient will affect the rationality and consistency of the result. In this research, we use a new type of Dempster-Shafer evidence theory to improve traditional DS/AHP. We have developed a prototype system for our algorithm and compared with the traditional AHP and DS/AHP. The results show that our new DS/AHP has reduced the conflict coefficient impacts and the number of pairwise criteria comparison. It can also decrease the probability of unreasonable decision suggestion.