An optimal evaluating method for uncertainty metrics in reliability based on uncertain data envelopment analysis

In reliability engineering, there always exist uncertainties caused by the scarcity of data and information. Various metrics are developed to measure reliability under uncertainty, including evidence-theory-based reliability, interval-analysis-based reliability, fuzzy-interval-analysis-based reliability, posbist reliability and belief reliability. As these five metrics are difficult to be fully understood, it is hard to select the most appropriate one for a specific condition. This paper will propose an uncertain evaluating model to conduct an objective evaluation on these uncertainty metrics. An evaluating index system is established from the aspects of capabilities and adaptabilities. Uncertainty theory is adopted to deal with subjective uncertainties in the quantification process of the index system. Then an evaluating method based on uncertain data envelopment analysis is proposed to provide decision-makers with a succinct result for a given operational context. Finally, the evaluating method is illustrated with a numerical example, which shows that final metric choices vary with different requirements.