关键词: 空间轴承, Kmedoids聚类, 特征频率比值, 参数融合, 故障诊断

Abstract: Bearings are the core components of space inertial actuators such as flywheel and Control Moment Gyros (CMG), and their operating status directly affect the performance and service life of the whole machine. At present, since that lightload bearings may produce a phenomenon like weak fault characteristics during normal operation, it is difficult to identify the normal and weak fault states for a single fault characteristic parameter. Aiming at this problem, this paper proposes a vibration fault clustering fusion method for bearing weak fault identification. First, the data are obtained through the bearing vibration experiment. Then, the characteristic parameters of the vibration signal are extracted based on the characteristic frequency ratio. On this basis, the KMedoids algorithm is used to cluster normal samples, and the safe boundary of normal operation is constructed according to the 3σ rule. Finally, the overrun probability of different bearing fault data are calculated, and the fault state is identified according to the probability. The results show that this method is feasible and effective for the identification of normal and weak faults of bearings.