中文摘要：

提出一种基于样本差异度的基本概率指派（BPA）生成方法.建立三角模糊数模型,根据所提出的差异度函数计算模型和待测样本的差异度,生成初始BPA.为了消除干扰影响,对初始BPA进行冲突阈值判别并进行相应的冲突消解,使得传感器在受到干扰等情况下也可输出合理的BPA.鸢尾花分类实验表明,所提出的方法简单实用,具有较强的干扰消除能力.

英文摘要：

A method of generating basic probability assignment（BPA） based on sample difference degree is proposed. The triangle fuzzy number models are established, and the proposed difference degree function is used to calculate the difference between the models and the sample under test to generate the initial BPA. In order to eliminate interference effects, the distance between the evidences is calculated, and conflicts are resolved according to the distance calculated and the threshold. In this way, it can output reasonable BPA under the condition of interference and so on. The iris data classification test shows that the proposed method is simple and practical, and has strong ability of interference elimination.