基于相对熵距离的动态改进理想解法及其应用研究  

Dynamic Improved Topsis Based on Relative Entropy Distance and Its Application Research

作　　者：李美娟 刘佳鸿 杨隆浩 胡慧芳 LI Meijuan;LIU Jiahong;YANG Longhao;HU Huifang(School of Economics and Management,Fuzhou University,Fuzhou 350108)

机构地区：[1]福州大学经济与管理学院,福州350108

出　　处：《系统科学与数学》2023年第1期174-185,共12页Journal of Systems Science and Mathematical Sciences

基　　金：国家自然科学基金项目(71872047,72001043,71702033);福建省百千万人才工程省级人选,福建省自然科学基金项目(2021J01570);福建省社科规划项目(FJ2020B028)。

摘　　要：为综合比较多个多指标对象在某时刻的发展状况和在不同时刻的整体发展态势,构建了基于相对熵距离的动态改进理想解法.该方法在传统理想解法基础上用改进熵值法确定不同时刻的指标权重,通过相对熵计算与理想解的距离避免了欧式距离的一些弊端,并增加考虑正负距离的相对重要性.利用基于波动性和时间度的时间权向量二次加权以推广到动态数据应用场景,最后通过实例验证该方法的可行性.In order to comprehensively compare the development status of multiple multi-index objects at a certain time and the overall development trend at different times,a dynamic improved TOPSIS based on relative entropy distance is constructed.Based on the traditional ideal solution,the improved entropy method is used to determine the index weight at different times.The distance between the relative entropy calculation and the ideal solution avoids some disadvantages of the Euclidean distance,and increases the relative importance of considering the positive and negative distance.The quadratic weighting of time weight vector based on volatility and time degree is used to extend to dynamic data application scenarios.Finally,the feasibility of this method is verified by an example.

关 键 词：动态评价方法 理想解法 相对熵 时间权重 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]