A reliability assessment approach for systems with heterogeneous component information

摘要

System reliability is assessed during development, production and delivery stages. Maintenance and warranty policies are decided based on this input. Reliability assessments are usually based on degradation and lifetime data treated separately. This article presents an approach to combine them using a comprehensive framework. Usually reliability assessment is done based on component-level tests and with information on system structure. Mostly point estimates are carried out on reliability, but some circumstances need interval estimation with a lower confidence limit as a proper assessment target. Among recent developments, the exact limits are preferred against Buehler limits which have application only in limited situations. Another method is an approximation that generates a balance between availability and accuracy. Simulation approaches have been proposed for reliability assessment, but they may produce inaccurate results if the sample size is small. As many complex systems are produced, reliability data has become more heterogeneous. In such cases degradation data is more reliable with limited lifetime data. However, compiling degradation and lifetime data is a difficult task. One method suggested in the literature is using a Bayesian approach. However, the priority nature of the Bayesian approach is not readily acceptable in this context since subjective biases caused by a priority nature of components with small sample sizes will get in the system level reliability data. To overcome these shortcomings, this article proposes a method to compile reliability data from two types of tests to generate a meaningful estimate of the system reliability.