%0 Journal Article
%A Zhao, Liang
%A Yang, Zhanping
%A Xiao, Longyuan
%D 2016
%I Begell House
%K uncertainty quantification, evidence theory, uncertain observation, kernel density estimation
%N 2
%P 157-165
%R 10.1615/Int.J.UncertaintyQuantification.2016016572
%T CONSTRUCTION OF EVIDENCE BODIES FROM UNCERTAIN OBSERVATIONS
%U http://dl.begellhouse.com/journals/52034eb04b657aea,3cc9ec274644f0dc,71dccf740cd59450.html
%V 6
%X The construction of evidence bodies is a key issue when the evidence theory is applied in uncertainty quantification. The existing approaches proposed for this topic are usually too subjective to obtain rational evidence bodies in the situation of uncertain observations. This paper introduces a repeated kernel-density-estimation based approach for constructing evidence bodies from uncertain observations. The typical uncertain observations-limited point measurements together with interval measurements are considered in this paper. Using kernel density estimation with a loop, a family of probability distribution about the given observations is obtained, the probability box characterized by the bounds of the probability distribution family is discretized to evidence bodies by an outer discretization method. The approach also considers the uncertainty in the distribution assumption during the kernel density estimation. A numerical example is used to demonstrate the proposed approach.
%8 2016-10-11