Reliability assessment based on D-S evidence theory

For many high-reliability products, reliability assessment relying on the tests has become unrealistic, but expert experience shows its advantage to the above problem. In order to resolve the problem that how to fuse the expert experience, we convert the expert experience into the reliability probability density distribution, determine the knowledge framework and basic probability assignment function based on the definition of consistent degree, fuse the basic probability distribution based on D-S fusion rule, and get the fusion distribution weights, finally obtain the reliability of the product. This method has been validated by an example, and shows perfect.

I. Introduction

In the process of assessing the reliability of the complex and high-reliability products, the traditional method based on mathematical statistic theories is no longer applicable to this situation, because of the limitation of the cost, time and so on. Reliability engineers, system designers and domain experts accumulate a large number of knowledge and experience that reflect the actual situation of the product, so expert experience should be given attention. Different experts have different decisions, so it is necessary to fuse different decisions in order to get higher precise and credible result [1, 2].