针对石油套管缺陷超声无损检测（NDT）中缺陷回波的特点，提出了一种基于小波包分解和支持向量机（SVM）的缺陷智能识别新方法。分析了Gabor、小波和小波包3种信号时频变换分解方法的特点，并进行了基于3种方法生成的特征数据可分性比较，确定了小波包分解方法效果最好。根据SVM解决分类问题的原理，采用SVM法对3种时频分解提取的缺陷信号特征数据进行识别。试验表明，基于小波包分解局部熵的特征提取结合SVM模式智能识别的组合方法，可应用于石油套管上的4种典型缺陷的识别。
A new intelligent flaws identification method was presented base on wavelet packet decomposition and support vector machine （SVM）, according to the characteristics of ultrasonic nondestructive testing （NDT） echo signals. The characteristics of Gabor transform, wavelet transform, and wavelet packet transform in signal de- composing were discussed. The separability of features achieved by three methods above was compared, and the wavelet packet method was proved to be the best. The classification principle of $VM method was introduced. And it was adapted to identify the features achieved by three time-frequency decomposing methods. The features extraction method with the SVM algorithm was proved to be efficient to identify four typical flaws in oil casing pipe.
参考文献
| [1] | Simonea G;Morabitoa F C;Polikarb R.Feature Extraction Techniques for Ultrasonic Signal Classification[J].Interna- tional Journal of Applied Electromagnetics and Mechanics,2002(15):291. |
| [2] | 刘清坤,阙沛文,宋寿鹏.基于支持向量机的石油管线缺陷识别方法研究[J].传感器技术,2005(03):30-31,34. |
| [3] | 陶卿,姚穗,范劲松,方廷健.一种新的机器学习算法：Support Vector Machines[J].模式识别与人工智能,2000(03):285-290. |
| [4] | Blanco S F;Igfiosa A;Quian Q R et al.Time-Frequency Analysis of Electroencephalogram Series(III):Information Transfer Func- tion and Wavelets Packets[J].Physical Review E,1998,57(01):932. |
| [5] | 边肇祺;张学工.模式识别[M].北京:清华大学出版社,2000 |
| [6] | 祁亨年.支持向量机及其应用研究综述[J].计算机工程,2004(10):6-9. |
| [7] | 刘志刚,李德仁,秦前清,史文中.支持向量机在多类分类问题中的推广[J].计算机工程与应用,2004(07):10-13,65. |
| [8] | 车红昆;项占琴;程耀东 .超声检测信号时频邻域白适应消噪技术[J].机械工程学报,2007,43(06):227. |
| [9] | 车红昆,吕福在,项占琴.多特征SVM-DS融合决策的缺陷识别[J].机械工程学报,2010(16):101-105. |
| [10] | Lfizaro J C;San Emeterio J L;Ramos A .Noise Reduction in Ultrasonic NDT Using Discrete Wavelet Transform Process- ing[J].IEEE Transactions on Ultrasonics Symposium,2002,43(05):777. |
- 下载量()
- 访问量()
- 您的评分：
-
10%
-
20%
-
30%
-
40%
-
50%