Refine your search:     
Report No.
 - 

Acoustic detection of bubble jets based on deep learning; Label classification and regression analysis

Okumura, Shuya*; Ueki, Yoshitaka*; Aizawa, Kosuke ; Ara, Kuniaki*

The Sodium-cooled Fast Reactor (SFR) is a next-generation nuclear reactor that uses liquid sodium as the reactor coolant. It is important to develop a method for early detection of bubble jets generated in the event of steam generator heat transfer tube failure. To solve this problem, we will demonstrate whether a more detailed understanding of jet condition is possible by performing regression analysis in addition to classification problems using time-frequency representation (TFR) and convolutional neural networks (CNN), as in previous studies. First, for proof of concept, pure water and air were used as the simulated fluids, and bubble jet sounds simulating abnormal sounds are obtained from three types of pipes with different diameters. Next, two types of TFRs were extracted from the acoustic data based on short-time Fourier transform and continuous wavelet transform. Finally, the TFRs were input to three different CNNs and trained. As a result, the model combining ResNet18 and STFT showed an accuracy of more than 99 % in label classification and a coefficient of determination of more than 0.99 in regression analysis. These results indicate that the proposed method based on deep learning has the potential to contribute to the understanding of bubble jet conditions in real SFRs.

Accesses

:

- Accesses

InCites™

:

Altmetrics

:

[CLARIVATE ANALYTICS], [WEB OF SCIENCE], [HIGHLY CITED PAPER & CUP LOGO] and [HOT PAPER & FIRE LOGO] are trademarks of Clarivate Analytics, and/or its affiliated company or companies, and used herein by permission and/or license.