한국음향학회지 (The Journal of the Acoustical Society of Korea)

  • 제42권4호
  • /
  • Pages.345-356
  • /
  • 2023
  • /
  • 1225-4428(pISSN)
  • /
  • 2287-3775(eISSN)
한국음향학회 (The Acoustical Society of Korea)
권호 표지
DOI QR코드

DOI QR Code

어텐션 기반 게이트 순환 유닛을 이용한 수동소나 신호분류

Passive sonar signal classification using attention based gated recurrent unit

  • 이기배 (제주대학교 해양시스템공학과) ;
  • 고건혁 (제주대학교 해양시스템공학과) ;
  • 이종현 (제주대학교 해양시스템공학과)
  • Kibae Lee ;
  • Guhn Hyeok Ko ;
  • Chong Hyun Lee (Department of Ocean System Engineering, Jeju National University)
  • 투고 : 2023.05.03
  • 심사 : 2023.06.23
  • 발행 : 2023.07.31
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

수동소나의 표적신호는 수초 내 세기의 변화를 갖는 협대역 고조파 특성과 로이드 거울 효과에 의한 장시간 주파수 변이 특성을 나타낸다. 본 논문에서는 지역 및 전역적 시계열 특징을 학습하는 게이트 순환 유닛 기반의 신호분류 알고리즘을 제안한다. 제안하는 알고리즘은 게이트 순환 유닛을 이용한 다층 네트워크를 구성하고 확장된 연결을 통해 지역 및 전역적 시계열 특징들을 추출한다. 이후 어텐션 메커니즘을 학습하여 시계열 특징들을 가중하고 수동소나 신호를 분류한다. 공개된 수중 음향 데이터를 이용한 실험에서 제안된 네트워크는 96.50 %의 우수한 분류 정확도를 보였다. 이러한 결과는 기존의 잔차 연결된 게이트 순환 유닛 네트워크과 비교하여 4.17 % 높은 분류 정확도를 갖는다.

Target signal of passive sonar shows narrow band harmonic characteristic with a variation in intensity within a few seconds and long term frequency variation due to the Lloyd's mirror effect. We propose a signal classification algorithm based on Gated Recurrent Unit (GRU) that learns local and global time series features. The algorithm proposed implements a multi layer network using GRU and extracts local and global time series features via dilated connections. We learns attention mechanism to weight time series features and classify passive sonar signals. In experiments using public underwater acoustic data, the proposed network showed superior classification accuracy of 96.50 %. This result is 4.17 % higher classification accuracy compared to existing skip connected GRU network.

키워드

과제정보

이 논문은 2023년도 정부(방위사업청)의 재원으로 국방기술진흥연구소의 지원을 받아 수행된 연구임(20-106-B00-003).

참고문헌

  1. Y. C. Jung, B. U. Kim, S. K. An, W. J. Seong, and K. H. Lee, "An algorithm for submarine passive sonar simulator" (in Korean), J. Acoust. Soc. Kr. 32, 472-483 (2013). https://doi.org/10.7776/ASK.2013.32.6.472
  2. K. Lee, J. Kim, M. J. Cheong, Y. W. Ryu, J. Bae, and C. H. Lee, "Separation and denoising algorithm of passive sonar target signals in frequency domain" (in Korean), J. IEIE, 55, 88-96 (2018). https://doi.org/10.5573/ieie.2018.55.3.88
  3. J. K. Ahn, H. D. Cho, D. Shin, T. Kwon, and G. T. Kim, "LOFAR/DEMON grams compression method for passive sonar" (in Korean), J. Acoust. Soc. Kr. 39, 38-46 (2020).
  4. S. E. Lee, S. B. Hwang, and D. Y. Noh, "A study on the algorithm for underwater target automatic classification using the passive sonar" (in Korean), J. KIMS Technol. 3, 76-84 (2000).
  5. B. Mishachandar and S. Vairamuthu, "Diverse ocean noise classification using deep learning," Appl. Acoust. 181, 108141 (2021).
  6. G. Song, X. Guo, W. Wang, Q. Ren, J. Li, and L. Ma, "A machine learning-based underwater noise classification method," Appl. Acoust. 184, 108333 (2021).
  7. C. Satheesh, S. Kamel, A. Mujeeb, and M. H. Supriya, "Passive sonar target classification using deep generative β-VAE," IEEE Signal Process. Lett. 28, 808-812 (2021). https://doi.org/10.1109/LSP.2021.3071255
  8. N. H. Bach, L. H. Vu, and V. D. Nguyen, "Classification of surface vehicle propeller cavitation noise using spectrogram processing in combination with convolution neural network," Sensors, 21, 3353 (2021).
  9. F. Liu, T. Shen, Z. Luo, D. Zhao, and S. Guo, "Underwater target recognition using convolutional recurrent neural networks with 3-D mel-spectrogram and data augmentation," Appl. Acoust. 178, 107989 (2021).
  10. S. Kim, S. K. Jung, D. Kang, M. Kim, and S. Cho, "Application of the artificial intelligence for automatic detection of shipping noise in shallow-water" (in Korean), J. Acoust. Soc. Kr. 39, 279-285 (2020).
  11. K. M. Park and D. Kim, "Preprocessing performance of convolutional neural networks according to characteristic of underwater targets" (in Korean), J. Acoust. Soc. Kr. 41, 629-636 (2022).
  12. A. Sherstinsky, "Fundamentals of recurrent neural network (RNN) and long short-term memory (LSTM) network," Physica D: Nonlinear Phenomena, 404, 132306 (2020).
  13. J. Chung, C. Gulcehre, K. H. Cho, and Y. Bengio, "Empirical evaluation of gated recurrent neural networks on sequence modeling," Proc. NIPS Workshop on Deep Learning, 1-9 (2014).
  14. X. Zhang, Y. Sun, K. Jiang, C. Li, L. Jiao, and H. Zhou, "Spatial sequential recurrent neural network for yperspectral image classifcation," IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 11, 4141-4155 (2018). https://doi.org/10.1109/JSTARS.2018.2844873
  15. S. Singh, S. K. Pandey, U. Pawar, and R. R. Janghel, "Classification of ECG arrhythmia using recurrent neural networks," Procedia Computer Science, 132, 1290-1297 (2018). https://doi.org/10.1016/j.procs.2018.05.045
  16. S. Chang, Y. Zhang, W. Han, M. Yu, X. Guo, W. Tan, X. Cui, M. Witbrock, M. Hasegawa-Johnson, and T. S. Huang, "Dilated recurrent neural networks," Proc. 31st Conf. NIPS, 1-13 (2017).
  17. S. H. Kang, "A study on the Lloyd's mirror effect on the underwater radiated noise for the underwater vehicle" (in Korean), J. Acoust. Soc. Kr. 40, 314-319 (2021).
  18. L. E. Kinsler, A. R. Frey, A. B. Coppens, and J. V. Sanders, Fundamentals of Acoustics (John Wiley & Sons, New Jersey, 1999), Chap. 15.
  19. D. S. Domingues, S. T. Guijarro, A. C. Lopez, and A. P. Gimenez, "ShipEar: An underwater vessel noise database," Appl. Acoust. 113, 64-69 (2016). https://doi.org/10.1016/j.apacoust.2016.06.008