A Novel Intelligent Method for Bearing Fault Diagnosis Based on Hermitian Scale-Energy Spectrum

Abstract

Early bearing fault diagnosis can prevent from production accidents and improve safety, so it has an important practical engineering significance. In this paper, an intelligent condition monitoring of rolling bearing based on Hermitian scale-energy spectrum (HSES) and time-domain characteristics is presented. First, aiming to the difficulty of feature extraction for rolling bearing, the HSES of continuous wavelet transform is proposed to analyze the vibration signal from the viewpoint of energy distribution. In order to build robust features, the dimensionless parameters are also used to assist in extracting time-domain characteristics. Then, these characteristics of the time and wavelet domains are mapped into a low-dimensional feature space by using the principal components analysis, which can reduce the redundancy of the features. Within the acquired feature space, support vector machine is used to intelligently determine the existence of bearing failure and classify the type of fault. Moreover, in order to make the model automatically optimize parameters in different working conditions, a designed genetic algorithm is implemented in this paper. Experimental results show that the proposed method can not only accurately recognize fault pattern but also effectively realize the automation during the whole process.