The diagnosis of satellite flywheel bearing cage fault based on two-step clustering of multiple acoustic parameters

Abstract

As the critical component of satellite flywheels, bearings should be tested and selected on the ground before the flight. Aiming at the condition assessment requirements of flywheel bearing cage in the ground test, this paper proposes a non-contact diagnosis method based on the acoustic signal. First, the acoustic signals of the flywheel bearing cage under normal, rubbing, and uneven lubrication conditions are acquired, and it is found that the traditional envelope demodulation method is difficult to distinguish these states effectively. Therefore, an idea based on multi-parameter clustering fusion diagnosis is proposed. Then, the limitations of the direct clustering method to identify and classify the types of cage faults are revealed. Finally, a new diagnosis and identification method based on two-step clustering is proposed to improve the diagnosis capability. The test results show that cage temporary instability and rubbing fault can be diagnosed and identified effectively by the proposed method.