Gear fault identification and classification of singular value decomposition based on Hilbert-Huang transform

An improved singular value decomposition method of gear fault identification based on Hilbert-Huang transform was proposed to overcome the problem of reconstructing a feature matrix of singular value decomposition. The method includes three steps. First, the instantaneous frequency and amplitude matrices were acquired by Hilbert-Huang transform from faulted gear signals. Second, after the matrices were decomposed by singular value decomposition, the defined distances of singular value vectors and the optimal threshold of the distance for classification were calculated. Third, the fault characteristics of a gearbox were identified and classified by the threshold of the distances. The result demonstrates that the proposed method effectively identifies the gear fault and can realize an automatic gear fault diagnosis.