RePofe: Reliability physics of failure estimation based on stochastic performance degradation for the momentum wheel

摘要

Momentum Wheel (MW) is the critical component for long life satellites. The reliability test, prediction and estimation of MW is of enormous challenges because of the special characteristics of small sample, long life and high test cost. Since it is impossible to obtain a large sample of failure data given limited money and time, the traditional statistical methods usually fail to resolve the reliability assessment problem of MW, and the efficient estimation of its reliability level can hardly be acquired. To address this problem, a novel reliability estimation framework in terms of physics of failure is advanced in this article, which is essentially based on the relationship between the MWs physics performance and its failure mechanisms. According to the actual physics of failure test and engineering analysis, a stochastic threshold Gauss-Brown process model is established to describe MW's failure process, and then a detailed parameter estimation method for the model is put forward. Finally, the effectiveness of our model is demonstrated through numerical computation based on the test data.