Fault Features Extraction and Identification based Rolling Bearing Fault Diagnosis

12th International Conference on Damage Assessment of Structures

Qin, B.^1 ; Sun, G.D.^1 ; Zhang, L.Y.^1 ; Wang, J.G.^1 ; Hu, J.^2

School of Mechanical Engineering, Inner Mongolia University of Science and Technology, Baotou

014010, China^1

Chungking Pure-Smart and Technology CO. LTD, Chungking

400030, China^2

关键词: Adaptive adjustment;    Classification accuracy;    Classification and identifications;    Extreme learning machine;    Fault classification;    Generalization ability;    High dimensional feature;    Permutation entropy;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/842/1/012055/pdf
DOI  :  10.1088/1742-6596/842/1/012055

来源: IOP

【 摘 要 】

For the fault classification model based on extreme learning machine (ELM), the diagnosis accuracy and stability of rolling bearing is greatly influenced by a critical parameter, which is the number of nodes in hidden layer of ELM. An adaptive adjustment strategy is proposed based on vibrational mode decomposition, permutation entropy, and nuclear kernel extreme learning machine to determine the tunable parameter. First, the vibration signals are measured and then decomposed into different fault feature models based on variation mode decomposition. Then, fault feature of each model is formed to a high dimensional feature vector set based on permutation entropy. Second, the ELM output function is expressed by the inner product of Gauss kernel function to adaptively determine the number of hidden layer nodes. Finally, the high dimension feature vector set is used as the input to establish the kernel ELM rolling bearing fault classification model, and the classification and identification of different fault states of rolling bearings are carried out. In comparison with the fault classification methods based on support vector machine and ELM, the experimental results show that the proposed method has higher classification accuracy and better generalization ability.