Bearing Fault Diagnosis Based on Improved Pseudo Fourthorder Moment
Zihan Wang, Wang Jian, Yongjian Sun · Journal of Artificial Intelligence Machine Learning and Data Science · 2023
In this paper, a new method of bearing fault diagnosis based on ensemble empirical mode decomposition (EEMD) and anchor point is advanced and verified.EEMD decomposition of vibration signals to rolling bearings on different working conditions, the pseudo-fourth-order moment (PFOM) of the decomposed signals calculated.Based on the definition of anchor point, a novel fault feature fitting anchor value proposed.Newton interpolation is adopted to fit the PFOM, and the anchor point is brought into the fitting function to get the anchor point fitting value.A large amount of data was obtained through many experiments, and the range of anchor fitting values under normal and five kinds of faults were determined.The anchor fitting value as the feature used to classify six working conditions.Taking the fitting value of the anchor point as the fault feature, the training data is selected and put into the extreme learning machine (ELM), and different working conditions are classified.Then take the test data and put it into the ELM model based on the training data for diagnosis, and the final classification accuracy reaches 97.5%.Finally, three comparative experiments shows that the present method are effective against the field of bearing fault diagnosis.