Reliability Modeling for Momentum Wheel Based on Data Mining of Failure-Physics

Reliability modeling has become an important issue in some special areas such as military and aerospace. Most existing reliability estimation methods are usually limited in these areas because of the requirement of large sample. Furthermore, the large sample test is impractical in terms of implementation expense and time, since most product have the characteristics of high reliability, high cost and small sample. All of these challenges motivate us to propose a new reliability estimation method which can be applied for the special areas. In this paper, we investigate the reliability estimation modeling based on small sample and data mining of failure-physics to address these challenges. Moreover, we demonstrate the performance gains of our method through physical experiment and reliability estimation on Momentum Wheel of satellite.