Application of Improved Back Propagation Neural Network for the Recognition of Composite Insulator Hydrophobicity Grade

Abstract:

The detection of hydrophobicity is an important way to evaluate the performance of composite insulator, which is helpful to the safe operation of composite insulator. In this paper, the image processing technology and Back Propagation neural network is introduced to recognize the composite insulator hydrophobicity grade. First, hydrophobic image is preprocessed by histogram equalization and adaptive median filter, then the image was segmented by Ostu threshold method, and four features associated with hydrophobicity are extracted. Finally, the improved Back Propagation neural network is adopted to recognize composite insulator hydrophobicity grade. The experimental results show that the improved Back Propagation neural network can accurately recognize the composite insulator hydrophobicity

• Periodical: Applied Mechanics and Materials (Volumes 373-375)
• Pages: 1155-1158
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.373-375.1155
• Online since: August 2013
• Authors: Kang Yan, Zhong Yuan Zhang
• Keywords: Adaptive Median Filter, Back Propagation (BP) Neural Network, Composite Insulator, Histogram Equalization, Ostu Threshold Method

References

[1] Song wei, et al: High Voltage Engineering, Vol. 3, (2005), pp.28-30. In Chinese.

[2] R. Barsch, H. Jahn, J. Lambrecht: IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 6, (1999), pp.668-675.

DOI: 10.1109/94.798123

[3] Lian Guan Kummel M.: Proc of SPIE, Vol. 6, (1989), pp.712-721.

[4] Gonzalez. R. C, Woods. R. E: Digital image processing, edited by Prentice-Hall, New Jersey (2000), pp.103-113.

[5] OTSUN: IEEE Trans, Vol. 9, (1979), pp.62-69.

[6] Martin T. Hagan, Mohammad B. Menhaj: IEEE Transactions on Neural Networks, Vol. 5, (1994), pp.536-538.