航天器飞轮轴承的性能退化评估与寿命预测研究

航天器飞轮 (反作用轮和动量轮)是航天器姿态控制系统的关键部件，其运转状态直接影响航天器在轨姿态控制能力。飞轮中的滚动轴承又是影响飞轮可靠性和制约航天器寿命的关键部件，因此，有必要对其进行性能退化评估和寿命预测。根据姿态控制的需要，航天器飞轮轴承常在变转速条件下工作，其次，当航天器经历日照区和阴影区时，还会产生较大的温度交变，进而导致轴承摩擦力矩（载荷）波动。上述的轴承工况变化给准确判断轴承状态带来影响，进而影响对轴承的性能退化评估与寿命预测。.本项目在动力学仿真和地面模型实验的基础上，研究轴承转速和温度变化对飞轮振动响应的影响，开展变工况条件下的特征提取，特征映射以及评估预测模型的研究，包括：变转速下飞轮振动参数化时频特征提取；基于冗余属性投影的变工况特征归一技术；基于动态贝叶斯网络的性能退化评估与寿命预测方法。本项目的研究成果将有效提高在轨航天器飞轮轴承性能退化评估和寿命预测的准确性。

Spacecraft flywheels (reaction wheels and momentum wheels) are key components of the attitude control system in spacecraft. The operation condition of flywheels can directly affect the attitude control ability for spacecraft. The rolling element bearing in spacecraft flywheel is a key part that can affect flywheel reliability and restrict spacecraft life. Therefore, it is necessary to carry out the research on performance degradation and life prediction of bearings in spacecraft flywheels. To fulfill the need of attitude control of spacecraft, flywheel bearings often work under varying rotating speed; at the same time, the alternating transition of spacecraft from the sunlight zone to the shadow zone causes fluctuation of friction torque in bearings. These varying operating conditions cause fluctuation of measured vibration signals, which causes difficulty in accurate performance assessment and life prediction of in-orbit bearings when directly using the measured vibration data..In this project, based on dynamic simulation and flywheel model experiment in ground, the influence of bearing speed and temperature variation on flywheel vibrations are studied. Feature extraction, feature mapping, assessment and prediction model under varying operating conditions will be investigated. The studies in this project include: parametric time-frequency feature extraction of flywheel vibration signals under varying rotating speed; feature normalization under varying working conditions based on nuisance attribute projection; performance degradation and life prediction of bearings in spacecraft flywheels based on dynamic Bayesian network. The study in this project will effectively increase the accuracy of the performance degradation and life prediction of in-orbit bearings of spacecraft flywheels.