The Indicator Variable Principal Component Neural Network Model for the Prediction of the Opening of Dam Crack

Abstract:

In order to effectively predict the opening of crack, the variation law of the crack is studied, and the indicator variable principal component neural network model is established by combining indicator variable model and principal component analysis with neural network theory. This new model can solve the problems of nonlinearity between the effect variable and the affecting factors and the colinearity between the affecting factors effectively and improved the prediction accuracy. By a case study, it showed that the prediction results are close to the actual values.

• Periodical: Applied Mechanics and Materials (Volumes 638-640)
• Pages: 774-777
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.638-640.774
• Online since: September 2014
• Authors: Jing Ji*
• Keywords: Crack Opening, DAM, Indicator Variable, Neural Network (NN), Principal Component

References

[1] Li Xuehong, Xu Hongzhong, Gu Chongshi, Zhou Xiaoxian, in: YANGTZE RIVER, Vol. 25, No. 5 (2001), pp.14-16.

[2] Zhen Dongjian, Gu Chongshi, Xu Shiyuan, in: dam observation and geotechnical tests, Vol. 25, No. 2 (2001), pp.28-30.

[3] Hagan M T., Demuth H B., Beale M H. Neural Network Design. BeiJing: China Machine Press(2002).