Effectiveness Verification of Bearing Health Assessment Method Based on SAE-MD

To verify the effectiveness of the health assessment model, experimental validation research on helicopter dynamic component was conducted. Firstly, the health representation of bearing signals was self-learning by stacked auto-encoder, and the health representation was input to Mahalanobis distance for health assessment. Then, the comprehensive design of test and verification profile for evaluating the degradation capability of bearing performance was carried out. Finally, the bearing degradation test and signal acquisition was carried out, and the collected vibration signals was analyzed and processed using the stacked auto encoder and Mahalanobis distance method, which verified the effectiveness of the health assessment methods proposed in this paper.