A Research On Reliability Evaluation Of High Reliability Product Based On Evidence Theory

The price of high reliability product tends to be expensive and it's not an easy question to recognize its faults.In the process of reliability assessment,high reliability product has some significant features which its test sample is less and test time is limited,as a result,we can obtain fewer reliability information from high reliability products than general products.The lack of experimental data and the promotion of product complexity make the probability evaluation method of reliability evaluation is no longer applicable.However,evidence theory can effectively make up for the lack of test samples by combining multi-source information and uncertainty can be well expressed and handled in evidence theory.Nowadays,the uncertainty evaluation method represented by evidence theory is a hot research area.So,in this article,we discuss a problem that how to apply evidence theory to evaluate reliability of high reliability product.To begin with,this paper proposes a new evidence combination rule based on the consistency among evidences,which completes efficient allocation of evidence conflict.Compared with traditional combination rules,the new rule has more accurate fusion results and it is more suitable for the fusion of high conflict evidences.Next,we present a general method of fault identification based on evidence fusion,evidence classification method is used to adjust failure belief assignment,which reduces the effect of interference information and achieve fast convergence of failure belief assignment.A fault diagnosis example combined motor rotor fault is given which verifies the feasibility of this method.In addition,we propose reliability assessment model based on evidence fusion.This method firstly defines the weights of information sources utilizing the relevant theories of cross entropy,and then amends the interval probability distribution of life values.Finally,we will get the life distribution of the product through transfer belief model and combination of the revise basic probability distribution,an example verifies the feasibility of the model.At last,the paper summarizes the research content in three aspects,are information fusion,fault diagnosis and life assessment.Subsequently we analyze the role of research results and further research was forecasted.