Flywheel bearing fault monitoring based on clustering fusion of normal operating acoustic parameter

The rolling bearings in flywheels are the core component of satellites. Its health condition plays a decisive role in the performance and the life of spacecraft. As the long-time test of flywheels on the ground, a convenient and reliable method for monitoring the operating state of abnormal bearings in flywheels is needed. Due to the unclear of fault mechanism and the insufficient of fault samples, a monitoring method based on the clustering fusion of normal operation acoustic parameter is proposed for abnormal of bearings. Firstly, the characteristics of flywheel's acoustic signals and its feasibility are clarified based on the tests. Then, statistical parameters and sound quality parameters are introduced to characterize the changes caused by bearing anomalies, and root mean square, roughness and sharpness are selected to construct the feature vectors. The K-medoids clustering technology is used to fuse the feature parameters, and the safe distance of normal operating bearings can be obtained. Finally, the safe distance is used to judge the bearing abnormal condition of several types of bearings through test.