Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 103
PROCEEDINGS OF THE THIRD INTERNATIONAL CONFERENCE ON SOFT COMPUTING TECHNOLOGY IN CIVIL, STRUCTURAL AND ENVIRONMENTAL ENGINEERING
Edited by: Y. Tsompanakis
Paper 13

Influence of Exceptional Data on Bridge Damage Assessment based on One-Class SVM Pattern Recognition

H. Furuta1, Y. Nomura2, K. Nakatsu3, H. Hattori4 and S. Yasuda5

1Faculty of Informatics, Kansai University, Japan
2Department of Science and Engineering, Ritsumeikan University, Japan
3Department of Modern Life, Osaka Jonan Women's Junior College, Japan
4Graduate School of Engineering, Kyoto University, Japan
5Graduate School of Informatics, Kansai University, Japan

Full Bibliographic Reference for this paper
H. Furuta, Y. Nomura, K. Nakatsu, H. Hattori, S. Yasuda, "Influence of Exceptional Data on Bridge Damage Assessment based on One-Class SVM Pattern Recognition", in Y. Tsompanakis, (Editor), "Proceedings of the Third International Conference on Soft Computing Technology in Civil, Structural and Environmental Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 13, 2013. doi:10.4203/ccp.103.13
Keywords: one-class support vector machine, neural network, pattern recognition.

Summary
Recently, it has become difficult to ensure the skill level of bridge maintenance engineers as a result of the retirement of experienced engineers. Hence, the daily inspection of bridges cannot be performed accurately and efficiently for the early detection of damage and to ensure appropriate measures. In previous research, damage assessment by introducing the technology of pattern recognition of digital images has been proposed. However, in the previous research, it was confirmed that there were some images which might be considered as exceptional data among the images of cracks of concrete decks. This paper describes the development of a support system for decision making in the damage assessment of bridges based on the pattern recognition by using the exception extraction. In this system, the exception extraction with one-class SVM removes some data that may cause the decrease of generalization capability of pattern recognition from the database. Thus, it may be expected that the accuracy and confidence of the results are improved. Numerical examples applying digital images of reinforced-concrete slabs of bridges with cracks are presented to demonstrate the practical use of the approach of this approach.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £42 +P&P)