A Method for Mass Estimation of Loose Parts in Nuclear Power Plant Based on Support Vector Machine

Abstract:

A new method for mass estimation of loose parts in nuclear power plant (NPP) based on the support vector machine (SVM) was proposed. It includes analyses of the relationship between the impact signals’ frequency spectrum and the mass of loose part, then formation of a vector consisting of linear predictive coding (LPC) parameters, which represent the shape of spectrum of impact signal. Using the vector as input data and the mass of loose part as the output data to train the SVM, the mass estimation can be done by the trained SVM model. Experimental results show that the method has higher accuracy and easier to achieve than the traditional methods. It provides a new way for mass estimation of loose part in NPP.

• Periodical: Key Engineering Materials (Volume 437)
• Pages: 384-388
• DOI: https://doi.org/10.4028/www.scientific.net/KEM.437.384
• Online since: May 2010
• Authors: Hua Wen Zheng, Yan Long Cao, Jiang Xin Yang, Yuan Feng He
• Keywords: Linear Predictive Coding, Loose Part, Mass Estimation, Support Vector Machine (SVM)

References

[1] B. Bechytold, U. Kunze: Progress in !uclear Energy. Vol. 34 (1999), p.221.

[2] W. Mayo: Progress in !uclear Energy. Vol. 34 (1999), p.263.

[3] S. Figedy, G. Oksa: Progress in !uclear Energy. Vol. 46 (2005), p.253.

[4] J. Makhoul: Proceedings of the IEEE (1975).

[5] L.B. Jackson: Digital Filters and Signal Processing, Second Edition (Kluwer Academic Publishers, 1989).

[6] Pan Xiang, LI Jiebing: Journal of Zhejiang University: Engineering Science (2006).

[7] V. Vapnik, S. Golowich, and A. Smola: Advances in !eural Information Processing Systems. 9, Cambridge, MA, MIT Press (1997).

[8] C. Cortes, V. Vapnik: Machine learning (1995).