| Status |
Accepted |
| journal_subject |
Part A |
| Article Type |
Regular Paper (More than 4 pages) |
| Article Filed |
test |
| Article Title |
Multi-fault Diagnosis for Roller Bearing Based on Extreme Learning Machine |
| Keywords |
extreme learning machine, roller bearing, fault diagnosis. |
| Abstract |
Single-hidden layer feed forward neural network (SLFN) for its good learning ability has been widely used in many fields. However, the inherent drawbacks such as slow learning speed and trivial tuned parameters become the main bottleneck restricting its development. It is clear that traditional algorithm can easily fall into local minimum points. For the requirements on rapidity and accuracy, extreme learning machine (ELM) is proposed in this paper for multi-fault diagnosis of rolling bearing. ELM is a new method based on SLFN, which provides better generalization performance at a much faster learning speed and with least human intervene. In this paper, time-domain and frequency-domain features are extracted from six different fault status of bearing, which can generally express the fault information of rolling bearing. And the classification model of ELM is established based on those feature set. The experiment results are compared with BP, RBF and SVM. It is obvious that ELM has a better effect even with a little number of neurons. Furthermore, the analysis of accuracy and effectiveness is discussed, showing that the algorithm proposed has better capacity of eneralization and a faster running speed. |
| Authors |
| First Name
| Middle Name
| Last Name
| E-Mail
| Corresponding
|
| Hongfang |
|
Yuan |
yuanhf@mail.buct.edu.cn |
Yes |
|
|
|
|
No |
|
|
|
|
No |
|
|
|
|
No |
|
| Upload Files |
Please Login to View Upload Files |
My Manuscript