Intelligent fault diagnosis for rotating machinery based on potential energy feature and adaptive transfer affinity propagation clustering

To identify fault types with a small amount of unlabeled fault data of rotating machinery, an intelligent fault diagnosis algorithm based on the potential energy feature and adaptive transfer affinity propagation clustering was proposed in this work. The algorithm can extract potential energy features from the intrinsic mode functions of a vibration signal using complete ensemble empirical mode decomposition with adaptive noise. An adaptive transfer judgment model is established from the source domain data after sensitive features extraction and self-weight analysis. The model can adjust the parameters according to the different target domains with unlabeled data. The effectiveness of the proposed intelligent fault diagnosis for roller bearings has been verified on different test-rigs, compared with the traditional classification techniques.