In view of the difficulty of extracting nonlinear and non-stationary vibration signals from planetary gearboxes

a planetary gearbox fault diagnosis method based on adaptive composite multi-scale permutation entropy (ACMPE)

improved supervised self-organizing incremental neural network landmark isometric mapping (ISSL-Isomap) and grey wolf optimizer support vector machine (GWO-SVM) was proposed. Fault features of vibration signals were extracted from the complex domain by using ACMPE

and the high-dimensional fault feature set was constructed. ISSL-Isomap was used to reduce the dimension of the high-dimensional fault feature set

and the low-dimensional and sensitive fault features were extracted. The low-dimensional fault features were input into a GWO-SVM classifier to recognize fault types. The analysis results of planetary gearbox fault diagnosis show that compared with the feature extraction methods of multi-scale permutation entropy (MPE) and composite MPE (CMPE)

ACMPE has more advantages in classification effect and recognition accuracy. ISSL-Isomap has the best dimensionality reduction effect compared with the dimensionality reduction algorithms of local tangent space alignment (LTSA)

isometric mapping (Isomap)

weighted Isomap (W-Isomap)

supervised Isomap (S-Isomap) and supervised self-organizing incremental neural network landmark Isomap (SSL-Isomap). The fault recognition rate of the proposed method reaches 100% with a certain superiority.