Gear Fault Diagnosis Method Based on Feature Fusion and SVM

Aiming at the gear fault diagnosis problems of the rotating machines, in order to improve the identification accuracy of the fault diagnosis and to extract the features that can more fully reflect the running state of the gear, a fault diagnosis method based on the feature fusion and support vector machine (SVM) was proposed. Firstly, using wavelet packet (WP), variational mode decomposition (VMD) and single energy (SE) extracted the features information of original vibration single respectively. Secondly, the extracted features were used to realize multiple groups linear feature fusion. Finally, the SVM classification method was adopted to evaluate the state of the running gear (normal, minor fault, medium fault, or broken tooth fault). Through experiment analyses and studies, it is shown that fusion features can reflect the running state of the gear more effectively and be helpful to achieve better diagnosis performance.