Fault Diagnosis Method of Bearings Based on Adaptive Time-Frequency Capsule Network
Bearing is a key component to ensure the safe operation of rotating machinery,and the traditional fault di-agnosis method is difficult to extract fault features and difficult to identify in the complex and changing operating environ-ment of bearings.An improved time-frequency adaptive capsule network-based bearing fault diagnosis method was pro-posed.Firstly,the one-dimensional original vibration signal was transformed into composite time-frequency structure da-ta by EEMD-HHT feature enhancement method to enhance the separability of non-stationary signal features.Then,the convolutional layer of the original capsule network was improved for adaptive depth extraction of time-frequency structure features of vibration signals.Finally,the capsule layer was introduced for the translation invariance of the convolutional neural network using dynamic routing algorithm to learn stored feature information for achieving intelligent diagnosis of fault types.The experimental results show that the proposed method has stronger fault-sensitive feature mining ability and higher diagnostic accuracy as well as working condition adaptive ability than the existing methods.
NETL
NSTL
万方数据
维普
