Service-Life Assessment of Complex Dynamic Systems Under Interval Uncertainty Based on Bayesian Networks

Service-life is a widely used reliability index in reliability engineering. For a complex dynamic system for which whole system tests are limited, and there is insufficient information to determine the distribution function of reliability models. Fortunately, the boundaries of lifetime variable can be obtained, which can be incorporated through the theory of interval uncertainty. In this study, a service-life assessment method for complex dynamic systems under interval uncertainty is introduced based on Bayesian networks (BN). Firstly, a dynamic fault tree (DFT) model is built for a system. Based on the comprehensive integration of test data, field data, design data and engineering experience, the lifetime of system units are expressed as interval numbers. Then, a coefficient of variation (COV) method is employed to determine the parameters of life distributions. Finally, the BN method is used to estimate the mean life of the example system, and the service-life of this system is assessed as well. The presented method can be easily used in engineering practice for service-life evaluation of complex dynamic systems under interval uncertainty, where lifetime data is limited.