Perturbation Analysis of Evidential Reasoning Rule
摄动(天文学)
庞加莱-林德斯特方法
计算机科学
应用数学
数学
物理
量子力学
作者
Shuaiwen Tang,Zhijie Zhou,Changhua Hu,Jianbo Yang,You Cao
出处
期刊:IEEE transactions on systems, man, and cybernetics [Institute of Electrical and Electronics Engineers] 日期:2019-10-15卷期号:51 (8): 4895-4910被引量:70
标识
DOI:10.1109/tsmc.2019.2944640
摘要
Evidential reasoning (ER) rule has been widely used in addressing uncertainty, ignorance, and vagueness information. To explore its performance measure (PM), the perturbation analysis (PA) for the ER rule (ER rule-PA) is conducted, with perturbation taken into consideration. This article aims to analyze the robustness and stability of the ER rule, serving as theoretical basis and technical support for applied research and applications. The combination of two pieces of independent evidence is discussed, and perturbation is added to one piece of evidence. To represent the expected utility of evidence combination under perturbation, perturbation utility is introduced. The novel concept of perturbation coefficient is proposed to characterize the PM of the ER rule (ER rule-PM). The properties of perturbation coefficient are explored to demonstrate the impact of perturbation. The maximum permissible error (MPE) of perturbation coefficient is defined to characterize the acceptability of perturbation. A numerical study is examined to illustrate the implementation process of ER rule-PA. Moreover, a case study of reliability evaluation of aerospace relay is conducted to show the potential applications of ER rule-PA, which makes the proposed method more practical.