한국음향학회지 (The Journal of the Acoustical Society of Korea)

  • 제38권1호
  • /
  • Pages.23-29
  • /
  • 2019
  • /
  • 1225-4428(pISSN)
  • /
  • 2287-3775(eISSN)
한국음향학회 (The Acoustical Society of Korea)
권호 표지
DOI QR코드

DOI QR Code

다중 주파수 대역 convolutional neural network 기반 지진 신호 검출 기법

Earthquake detection based on convolutional neural network using multi-band frequency signals

  • 김승일 (고려대학교 전기전자 공학부) ;
  • 김동현 (고려대학교 전기전자 공학부) ;
  • 신현학 (고려대학교 전기전자 공학부) ;
  • 구본화 (고려대학교 전기전자 공학부) ;
  • 고한석 (고려대학교 전기전자 공학부)
  • 투고 : 2018.11.29
  • 심사 : 2019.01.23
  • 발행 : 2019.01.31
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

본 논문에서는 국내에서 발생한 지진 신호를 검출 및 식별하기 위한 방법을 다루었다. 국내에서 발생한 지진 신호들을 분석해 본 결과 서로 다른 주파수 대역 신호의 특징들이 각각 분류를 위한 특징으로 적절함을 확인할 수 있었다. 이러한 분석 결과를 바탕으로 지진 신호에서 추출한 다중 주파수 대역 특징을 기반으로 하는 CNN(Convolutional Neural Network) 기법에 대해서 제안하였다. 제안하는 다중 주파수 대역 CNN 기법은 지진 신호에서 추출한 멜 스펙트럼에 대해서 각각 필터를 적용하여 서로 다른 주파수 대역(저/중/고 주파수)의 신호를 추출하였다. 추출된 신호들을 바탕으로 각각 CNN 기반 분류를 수행하였고, 수행된 결과를 융합하여 최종적으로 지진 이벤트에 대해 식별하였다. 2018년 동안 대한민국에서 발생한 실제 지진데이터를 기반으로 하는 실험을 통해 제안하는 기법에 대한 효용성을 검증하였다.

In this paper, a deep learning-based detection and classification using multi-band frequency signals is presented for detecting earthquakes prevalent in Korea. Based on an analysis of the previous earthquakes in Korea, it is observed that multi-band signals are appropriate for classifying earthquake signals. Therefore, in this paper, we propose a deep CNN (Convolutional Neural Network) using multi-band signals as training data. The proposed algorithm extracts the multi-band signals (Low/Medium/High frequency) by applying band pass filters to mel-spectrum of earthquake signals. Then, we construct three CNN architecture pipelines for extracting features and classifying the earthquake signals by a late fusion of the three CNNs. We validate effectiveness of the proposed method by performing various experiments for classifying the domestic earthquake signals detected in 2018.

키워드

GOHHBH_2019_v38n1_23_f0001.png 이미지

Fig. 1. Frequency average histogram: (a) strong earth-quake, (b) weak earthquake, (c) artificial (man-made) earthquake, (d) noise.

GOHHBH_2019_v38n1_23_f0002.png 이미지

Fig. 2. 2D visualization: (a) all frequency band, (b) low frequency band, (c) middle frequency band, (d) high frequency band. (red dot: strong earthquake, green dot: weak earthquake, blue dot: artificial (man-made) earthquake, magenta dot: noise/clutter).

GOHHBH_2019_v38n1_23_f0003.png 이미지

Fig. 3. Proposed CNN structure based analysis and learning of multi-band frequency characteristics of earthquakes.

GOHHBH_2019_v38n1_23_f0004.png 이미지

Fig. 4. Database examples: (a) strong earthquake, (b) weak earthquake, (c) artificial earthquake, (d) noise.

GOHHBH_2019_v38n1_23_f0005.png 이미지

Fig. 5. Confusion matrix of earthquake classification results: (a) Alexnet based baseline, (b) multi-band (proposed).

참고문헌

  1. M. Baer and U. Kardolfer, "An automatic phase picker for local and teleseismic events," Bull, Seismol. Soc. Amer. 1437-1445 (1987).
  2. C. D. Saragiotis and S. M. Panas, "PAI-SK: a robust automatic seismic P phase arrival identification scheme," IEEE Transactions on Geoscience and Remote Sensing, 1395-1395 (2002).
  3. R. J. Skoumal, M. R. Brudzinski, B. S. Currie, and J. Levy, "Optimizing multi-station earthquake template matching through re-examination of the Youngstown, Ohio, sequence," Earth Planet. Sci. Letter. 274-280 (2014).
  4. R. S. Jia, H. -M. Sun, Y. -J. Peng, and Y. -Q. Liang, "Automatic event detection in low SNR microseismic signals based on multi-scale permutation entropy and a support vector machine," J. seismology, 735-748 (2016).
  5. P. Singh, I. G. Roy, S. Kumar, and J. R. Kayal, "Seismic source characteristics in Kachchh and Saurashtra regions of Western India: b-value and fractal dimension mapping of aftershock sequences," Natural Hazards, 33-49 (2015).
  6. W. Y. Yun, S. -C. Park, and K. Y. Kim, "Comparison of background noise characteristics between surface and borehole station of Hwacheon" (in Korean), Geophysics and Geophysical Exploration, 203-210 (2013).
  7. E. Budakoğlu and G. Horasan, "Classification of seismic events using linear discriminant function (LDF) in the Sakarya region, Turkey," Acta Geophysica, 895-906 (2018).
  8. N. Jana, C. Singh, R. Biswas, N. Grewal, and A. Singh, "Seismic noise analysis of broadband stations in the Eastern Ghat Mobile Belt of India using power spectral density," Geomatics, Natural Hazards and Risk, 1947-5705 (2017).
  9. S. Mun, M. Shin, S. Shon, W. Kim, D. K. Han, and H. Ko, "DNN transfer learning based non-linear feature extraction for acoustic event classification," IEICE trans. on Information and Systems, 2249-2252 (2017).
  10. S. Mun, S. Shon, W. Kim, D. K. Han, and H. Ko, "A novel discriminative feature extraction for acoustic scene classification using RNN based source separation," IEICE Trans. on Information and System, 3041-3044 (2017).
  11. M. Curilem, J. P. Canario, and R. A. Rios, "Using CNN To classify spectrograms of seismic events from llaima volcano (Chile)" International Joint Conference on Neural Networks (IJCNN) (2018).
  12. T. Perol, M. gharbi, and M. Denolle, "Convolutional neural network for earthquake detection and location," Sci. Adv. 4, e1700578 (2018). https://doi.org/10.1126/sciadv.1700578
  13. W. Zhu and G. C. Beroza, "PhaseNet: A deep-neuralnetwork- based seismic arrival time picking method," arXiv:1803.03211 (2018).
  14. National Earthquake Comprehensive Information System (NECIS) Event wave downloads, http://necis.kma.go.kr/
  15. A. Krizhevsku, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," in Advances in Neural Information Processing Systems, 1097-1105 (2012).
  16. L. Wan, M. Zeiler, S. Zhang, Y. L. Cun, and R. Fergus, "Regularization of neural networks using dropconnect," Proc. Machine Learning, 1058-1066 (2013).