Keywords: high-strength bolt, axial force, pattern recognition, random forest, fast Fourier transform, attenuation rate.

Several structures were built during the high economic growth period in Japan, and bridges accounted for nearly forty percent of them. As these bridges have become relatively old, deterioration of the high-strength bolts used in their construction has become a major concern. Periodic diagnoses are essential to detect high strength bolts that have deteriorated because of corrosion, metal fatigue, and so on. The impact acoustics method is one of the methods used for diagnosing the axial force of high strength bolts. The impact acoustics method involves inspecting an object by listening and perceiving possible issues based on the impact sound; this method is usually performed by skilled engineers and it is preferred for bridge diagnosis because of its advantages of low risk, low cost, ease of implementation, and high reliability.

However, the impact acoustics method is heavily dependent on experienced and skilled engineers; therefore, an alternative method that can accurately and efficiently diagnose the condition of high strength bolts is required. To solve these problems, we proposed an unsupervised system for assessing the axial force of high strength bolts. In the proposed system, we hit a high-strength bolt with a hammer, and obtain the frequency data from the waveform data of the impact sound. Furthermore, we obtain the attenuation rate from the original sound waveform data. Then we use the random forest classifier to diagnose the axial force of the high-strength bolts. Finally, we compared the recognition accuracy of the proposed system to verify its efficacy.

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 £75 +P&P)