A Hydraulic Fault Diagnosis Method Based on IMF Entropy Feature Fusion

Abstract

A feature extraction and fault diagnosis method based on IMF entropy feature fusion was proposed for external vibration signals in five common fault states of hydraulic equipment: normal, leakage, blockage, air cavity and impact. Firstly, all kinds of signals were decomposed by improved EMD based on frequency cutoff, and effective IMF components were screened, then the fusion features of multiple information entropy were extracted, and then the deep learning method of DBN was adopted for feature learning and status recognition. The experimental results show that this method has high recognition accuracy and can effectively realize multi-fault recognition of hydraulic system.