Combined Improved EEMD with SVM in the Bearing Low Dimensional Small Sample Fault Diagnosis

Abstract:

In order to diagnose bearing fault with less samples,combined improved EEMD with SVM in the bearing fault intelligent diagnosis under low dimensional small sample is researched in this paper.It is applied to the binary classification and identification in bearing normal and fault state.The results show that depend only on less sample data 5d feature vector classification after training, SVM using linear kernel function and polynomial kernel function classification accuracy is still up to 100. classification accuracy under the less sample data in less 5d characteristic vectors by RBF kernel function under Sigmoid kernel function is relatively low.Choose appropriate SVM kernel function completely can realize low dimensional small sample right binary classification.

• Periodical: Applied Mechanics and Materials (Volumes 427-429)
• Pages: 354-357
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.427-429.354
• Online since: September 2013
• Authors: Mei Jun Zhang*, Kai Chai, Jie Huang, Hao Chen
• Keywords: Classification, EEMD, Identification, Intelligent Diagnosis, Low Dimensional Small Sample, Support Vector Machine (SVM)

References

[1] Hwi Li, Xiaofeng Liu and Lin Bo. Journal of Vibration, Measurement & Diagnosis In Chinese. Vol. 31(2011) , pp.591-595.

[2] Jiangcang Ma, Jing Chen and Xiaolong Liu. Computer Measurement & Control In Chinese. Vol. 17(2009), pp.2115-2117.

[3] N. E Huang, Z Shen and S. R Long. Proceeding of the Royal Society, London, Vol. 454(1998), pp.903-995.

[4] Zhaohua Wu and Norden E. Huang. Advances in Adaptive Data Analysis，Vol. 1(2009), pp.1-41.

[5] Meijun Zhang, Sichen Han, Chuang Wang and Shuguang Li. Journal of Advanced Materials Research. Vol. 479-481(2012) , pp.1180-1185.

[6] Zhang Meijun , Chen Hao , Cao Qin and Chuang Wang . Journal of Vibration, Measurement & Diagnosis in Chinese. Vol. 33(2013), pp.93-98.