Abstract: The orbit environment of a satellite momentum wheel is very complicated, and there is hardly any failure data for it. Therefore it is nearly impossible to model and analyze its reliability using conventional methods. In view of this, the article presents a momentum wheel reliability modeling and assessing method using Bayesian network through fusing various test information and telemetric data such as bearing temperature and electric current. The topology of Bayesian network is founded based on failure analysis and the parameter is estimated using test data. Then, the article carrys out point estimate and confidence interval estimate of reliability for a momentum wheel using the inference of Bayesian network and assesses the real time reliability based on telemetric data. And then, the telemetric data interval in which momentum wheel’s reliability is very sensitive is obtained. The method is important for the reliability real time observing of momentum wheel in theory and practice.

Key words: reliability, Bayesian network, telemetric data, momentum wheel, orbit environment