Abstract: The momentum wheel is one of the most important electromechanical components of a satellite pose control system, and its reliability estimation is a key issue in satellite launch. Limited by technology, cost, and time etc., it is very difficult to model, analyze, and estimate the reliability of a momentum wheel by means of traditional methods, since momentum wheels are characterized by small sample size, high reliability requirements, and long life, and it is hard to get large samples of failurelife data in a short time. In view of the problem, this article first establishes a performance degradation model based on failure physics, and then integrates the performance degradation and life models to obtain a reliability model, and uses the performance data and scanty life data to estimate the reliability based on the new model. The instance analysis proves that the failure-physics-analysis-based method of Bayesian reliability estimation for the momentum wheel is more effective and consistent with engineering facts than the fakelife-based method.

Key words: Bayesian method, failure physics, small sample, momentum wheel, performance degradation, reliability