基于D-S证据理论的灰色定权聚类综合后评价方法

Method of grey fixed weight clustering comprehensive ex-post evaluation based on D-S evidential theory

结合灰色定权聚类评估模型和D-S证据理论对项目综合后评价方法进行研究.应用差异信息序列熵理论,根据指标所蕴含信息量的多少确定聚类指标的权重;根据所得到的灰色定权聚类系数矩阵并把该矩阵经过适当转换,把每一个聚类对象作为影响后评价结果的一条证据,考虑到证据推理中零绝对化问题存在的可能性,给出了解决办法;利用Dempster合成法则,得到了辨识框架中各子集的信度函数,根据信度函数最大值确定项目综合后评价的结果.可以充分利用定权聚类中所得到各聚类对象所属灰类的信息,减少了直接使用聚类系数最大化方法确定评价对象所属灰类造成的信息损失,最后通过一个算例说明了该方法的实用性和有效性。

This paper studies a method of comprehension ex-post evaluation of projects by the application of grey clustering evaluation model and Dempster-shafer evidential theory. Such theory as information entropy of difference information sequences is applied in the paper, by which weights of clustering indices are assigned according to information volume. Under conversions of grey fixed weight clustering coefficients, ones look every clustering object as an evidence impacting on the result of ex-post evaluation and consider probable existence of zero absolutiation problem, a method is given to solve it. Then ones get belief functions of subsets in frame of discernment with application of Dempster combination rule and the result of evaluation in terms of maximum value of belief functions. After the fusion of data which come from grey fixed clustering coefficients matrix, ones could apply information of grey category of each clustering object fully. This method also decreases information loss than the directly application of maximum value method of grey clustering coefficients. Finally, an example is presented to illustrate the practicality and effectiveness of the method mentioned in this paper.