Advances in data combination, analysis and collection for system reliability assessment

摘要

Reliability is defined as the probability a system will perform its intended function for at least a given period of time when operated under some specified conditions. However, when the systems are complex in nature, such as nuclear weapons, infrastructure networks, supercomputer codes and munitions, the system reliability methodology is faced with several challenges. The present research work focuses on four methodological issues that arise from complex systems reliability problems. In Section 2 of the paper, the methods for integrating multiple data sources to assess the reliability of a single component are presented. The data may come from many sources, including experimental test results, computer simulation, and expert opinion. In this section, the researchers have considered specifically degradation data, surrogate data, and a biased sample of pass/fail data. In Section 3, the methods for assessing system reliability when the data are available at multiple levels is considered. The researchers have considered situations where the data is available about both components and the combination of the components. Examples of combining failure time data, failure count data, Bernoulli data, and degradation are discussed. In Section 4, the Bayesian networks and flow graph models are discussed, for Bayesian networks generalize fault trees and reliability block diagrams and flow graph models are multistate models that simplify the analysis of time-to-event data. In Section 5 of the paper, the researchers present the resource allocation wherein the focus is on addressing how to allocate limited testing resources. In Section 5, the resource allocation problems for systems are considered. Section 6 summarizes the view of the current research challenges in systems reliability assessment.