一种新的解决冲突问题的不确定性度量方法

自动化数据处理、检测判决、推理决策要求对多种传感器和不同信息源进行有效集成.但由于环境的扰动、传感器的局限以及人为的干预,信源信息一般具有较强的不确定性、不完备性、冲突性,集中表现为证据的冲突,有必要对冲突证据的推理与决策进行研究.为解决证据理论不能有效融合冲突证据的问题,提出一种新的不确定性度量方法.首先,对现有的基于信息熵和区间距离的不确定性度量公式进行深入分析,总结现有方法的缺陷并进行相关证明;其次,基于定积分定义新的区间距离度量公式,并对公式的合理性进行证明;然后,基于所提出的区间距离公式给出新的不确定性度量方法,利用改进的不确定性度量方法给出冲突证据组合规则及算法流程;最后,通过算例分析验证改进算法的有效性和可行性.

A new uncertainty measurement method for conflict problem

Automated data processing, detection decisions, and inference decisions require efficient integration of multiple sensors and different information sources. However, due to environmental disturbances, sensor limitations, and human intervention, source information generally has strong uncertainty, incompleteness, and conflict, and is concentrated in evidence conflicts. Therefore, it is necessary to study the reasoning and decision-making of conflict evidence. In order to solve the problem that evidence theory cannot integrate conflict evidence effectively, a new uncertainty measurement method is proposed in this paper. First of all, through an in-depth analysis of the existing uncertainty measurement formulas based on information entropy and interval distance, the defects of the existing algorithms are summarized and proved. Then based on definite integral, a new interval distance measurement formula is defined and the rationality of this formula is proved. Furthermore, based on the proposed interval distance formula, a new uncertainty measurement method is presented and the conflict evidence combination rule and algorithm flow are given using the improved uncertainty measurement method. Finally, the feasibility and effectiveness of the improved method are illustrated by an example.