基于马田系统和模糊积分的多属性决策方法

发布时间：2018-03-18 03:24

本文选题：马田系统　切入点：模糊测度　出处：《管理工程学报》2015年03期 　论文类型：期刊论文

【摘要】：在实际决策问题中,由于属性之间往往存在一定的交互作用,而传统的多属性决策方法都是建立在属性之间相互独立的假设前提之上,不能有效处理属性间存在的交互作用。针对这一问题,提出了一种基于马田系统和Choquet模糊积分的多属性决策方法。该决策方法首先根据马田系统的降维原理测度单个属性的信息增益,然后利用Sigmoid函数将之转化为单个属性的测度密度,进而求出能够描述属性间交互作用的模糊测度,最后基于所得到的模糊测度,利用Choquet模糊积分算子对各候选方案的评价信息进行集成。实例计算表明,利用本文方法确定的模糊测度具有更小的均方根误差,有助于提高决策的准确性,可以有效处理实际决策问题。
[Abstract]:In practical decision making problems, because of the interaction between attributes, the traditional multi-attribute decision-making methods are based on the assumption that attributes are independent of each other. Cannot effectively handle the interaction between attributes. To address this problem, In this paper, a multi-attribute decision making method based on Multi attribute and Choquet fuzzy integral is proposed. Firstly, the information gain of a single attribute is measured according to the dimensionality reduction principle of the Matton system. Then it is transformed into the measure density of single attribute by Sigmoid function, and the fuzzy measure which can describe the interaction between attributes is obtained. Finally, based on the fuzzy measure obtained, The evaluation information of each candidate scheme is integrated with Choquet fuzzy integral operator. The example shows that the fuzzy measure determined by this method has smaller root mean square error, which is helpful to improve the accuracy of decision making. It can deal with practical decision making problems effectively.
【作者单位】：南京理工大学经济管理学院;安徽工业大学经济学院;
【基金】：国家自然科学基金资助项目(71271114,11071120) 教育部人文社会科学规划基金资助项目(10YJA630020)
【分类号】：C934

【参考文献】