|
| Adaptive Feature Extraction of Gear Fault Based on GWO-VMD Algorithm |
| |
| View Full Text View/Add Comment Download reader |
| DOI：10.7643/issn.1672-9242.2023.02.016 |
| KeyWord:gear acoustic emission signal variational mode decomposition grey wolf optimization peakedness sample entropy support vector machine |
| Author | Institution |
| CUI Le-han |
Shenyang University of Technology, Shenyang , China |
| YU Yang |
Shenyang University of Technology, Shenyang , China |
|
| Hits: |
| Download times: |
| Abstract: |
| The work aims to make use of the acoustic emission signal when fault occurs for adaptive feature extraction and diagnosis. The variational mode decomposition method (VMD) was used to decompose the acoustic transmission signal when fault occurs. In reality, collecting acoustic signal noise interference causes low feature extraction accuracy due to the large original noise interference, and mode decomposition parameters need artificial debugging set. The Gray Wolf Optimization (GWO) algorithm was introduced to obtain the intrinsic mode function (IMF) through signal decomposition after adaptive selection of optimal parameters for mode decomposition k and quadratic penalty factor α. The peakedness and sample entropy were calculated by selecting the based on correlation coefficient. The correlation coefficient of each component was calculated. The peakedness and the sample entropy of the component most similar to the original signal were calculated respectively. After the decomposition, the gear fault acoustic emission signal was higher than normal, while the sample entropy was more than the normal value under the fault condition. Support vector machine is used to classify and identify feature vector sets. Compared with the improved experimental results, GWO-VMD combined with peakedness-sample entropy method can effectively extract fault features and determine whether the gear state is healthy. |
| Close |