Gearbox fault diagnosis based on improved multi-scale fluctuation dispersion entropy and multi-cluster feature selection

Abstract

The vibration signal of a gearbox contains a large amount of information and can be used for fault diagnosis of gearboxes. In order to efficiently extract fault features from the vibration signals and improve the reliability of fault diagnosis, a gearbox fault diagnosis method based on improved multi-scale fluctuation dispersion entropy (IMFDE) is proposed. The method takes full advantage of sliding coarse-grained processing to alleviate the shortcomings of traditional multi-scale entropy methods and improve the stability of multi-scale fluctuating dispersion entropy (MFDE). The multi-cluster feature selection (MCFS) method is then combined with the selection of low-dimensional sensitive features from the original multi-scale features, and the sensitive feature matrix is input to a random forest (RF) classifier to mine the complex mapping relationship between the input features and the fault type to achieve fault diagnosis of gearboxes. Finally, experimental data of two gearboxes are used to verify the reliability of the proposed method. The results show that the proposed method can accurately determine different fault types of gearboxes and has significant advantages in terms of reliability and stability of fault identification compared with other existing methods.