臺灣博碩士論文加值系統

台灣位於歐亞板塊以及菲律賓海板塊之間，地震在台灣是相當常見的自然災
害，我們先前的研究已經在台灣佈署超過600 個以上的地震測站(Palert)。當地震
發生時，以偵測地震的初波P 波來進行預警，降低災害所帶來的損失。以往的做
法是利用STA/LTA 觸發器以及經過人工參數的判斷來偵測P 波的抵達，通過判
斷後再送入TCPD 進行fusion 端的地震計算。然而卡車的經過、施工事件或其他
導致地表震動的雜訊大部份依然會通過local 端參數的判斷，也會被送入TCPD
進行決策。這會影響在TCPD 計算地震時候的準確度。人工固定的參數並不能很
有效的解決雜訊的干擾，只能不停的修正再修正，來保持著可容忍的錯誤。在
這篇論文中我們利用了機器學習的方式來加強local 端的偵測。我們根據中央氣
象局(CWB) 專家的專業擷取在波形中的ㄧ些地震特徵值來進行KNN(K-nearest
neighbors) 學習、決策樹(Decision Tree) 學習以及支持向量機SVM(Support vector
machine) 學習。做交叉分析並比較其結果並選出較好的學習參數，比較模組的
f-score 值以及預測時間，最後選擇出其中表現最好的模組以上線進行地震的預
警。由於低成本的測站有利於廣泛的佈署，因此我們很容易取得大量的未發生地
震的資料產生，訓練資料集的內容可以包含幾乎大部分的雜訊情況。而台灣一年
有著上百個大小地震，地震資料也相當的豐富。

Since the dramatic changes in the global climate, earthquakes occurred more and
more frequently in the recent years. The damages caused by earthquakes have become
serious threats to human life and property. Providing early warning in realtime can be
a potential way to prevent casualties when massive earthquakes occur. However, false
alarms could also cause large loss or panic among people. Therefore, correctly detecting
the occurrence of an earthquake is a critical issue in disaster prevention and control. Since
signals collected by seismic sensors are usually corrupted by noise in the practice, many
detecting schemes have been proposed to provide better accuracy for earthquake detection.
Traditional schemes usually depend on manually selecting parameters or thresholds
in the detection process in order to reduce false alarms. The quality of detection decisions
is highly correlated to the experiences of the parameter and threshold selection. In
this paper, advanced machine learning mechanisms are adopted to learn from historical
events and identify the occurrence of earthquakes. Rather than relying on the experiences
of experts in geology, it can have objective and fair criteria in identifying the occurrence
of earthquakes and prevent false alarms caused by the experiences of human beings. The
seismic data collected by an experimented strong motion detection network built in Taiwan
are used to train the machine learning mechanisms including the K-Nearest Neighbor
(KNN) method, Classification Tree, and Support Vector Machine (SVM). Comparing to
the traditional detection method which selects thresholds based on experiences for certain
criteria to identify the occurrence of earthquakes, the proposed learning detection methods
performs much better in terms of detection accuracy. False alarms can be almost totally
prevented for large earthquake events.

Abstract in Chinese . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Abstract in English . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1 Earthquake Early Warning . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Anomaly Detection Methods . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Machine Learning in Seismic Applications . . . . . . . . . . . . . . . . . 5
2.4 Community Sensing Network . . . . . . . . . . . . . . . . . . . . . . . . 6
3 Earthquake Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.1 Seismic Motion Detection Network . . . . . . . . . . . . . . . . . . . . 7
3.1.1 Palert . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.1.2 Deployment of the Network . . . . . . . . . . . . . . . . . . . . 8
3.1.3 System Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2 Event screening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
v
3.3 Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.4 Event Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.4.1 K-Nearest Neighbors . . . . . . . . . . . . . . . . . . . . . . . . 17
3.4.2 Decision Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.4.3 Support Vector Machine(SVM) . . . . . . . . . . . . . . . . . . 24
4 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.1 Evaluation Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.2 Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.2.1 Events Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.2.2 Non-Earthquake Data . . . . . . . . . . . . . . . . . . . . . . . . 29
4.2.3 Example of the Data . . . . . . . . . . . . . . . . . . . . . . . . 29
4.3 Evaluation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.3.1 Cross Validation . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.3.2 K-Nearest Neighbors . . . . . . . . . . . . . . . . . . . . . . . . 33
4.3.3 Decision Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.3.4 Support Vector Machine . . . . . . . . . . . . . . . . . . . . . . 38
4.3.5 Heuristic Method . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.4 Comparison of the Methods . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.5 Computation Time for Prediction . . . . . . . . . . . . . . . . . . . . . . 49
5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Letter of Authority . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

[1] T. Kanamori, S.-I. Nishikawa, I. Shin, P. G. Schultz, and T. Endo, “Probing the environment along the
protein import pathways in yeast mitochondria by site-specific photocrosslinking,” Proceedings of the
National Academy of Sciences, vol. 94, no. 2, pp. 485–490, 1997.
[2] R. M. Allen, P. Gasparini, O. Kamigaichi, and M. Böse, “The status of earthquake early warning around
the world: An introductory overview,” Seismological Research Letters, vol. 80, no. 5, pp. 682–693,
2009.
[3] C. Satriano, Y.-M. Wu, A. Zollo, and H. Kanamori, “Earthquake early warning: Concepts, methods
and physical grounds,” Soil Dynamics and Earthquake Engineering, vol. 31, no. 2, pp. 106–118, 2011.
[4] O. Kamigaichi, M. Saito, K. Doi, T. Matsumori, S. Tsukada, K. Takeda, T. Shimoyama, K. Nakamura,
M. Kiyomoto, and Y. Watanabe, “Earthquake early warning in japan: Warning the general public and
future prospects,” Seismological Research Letters, vol. 80, no. 5, pp. 717–726, 2009.
[5] Y.-M. Wu and T.-l. Teng, “A virtual subnetwork approach to earthquake early warning,” Bulletin of
the Seismological Society of America, vol. 92, no. 5, pp. 2008–2018, 2002.
[6] Y.-M. Wu, D.-Y. Chen, T.-L. Lin, C.-Y. Hsieh, T.-L. Chin, W.-Y. Chang, W.-S. Li, and S.-H. Ker,
“A high-density seismic network for earthquake early warning in taiwan based on low cost sensors,”
Seismological Research Letters, vol. 84, no. 6, pp. 1048–1054, 2013.
[7] C. D. Saragiotis, L. J. Hadjileontiadis, and S. M. Panas, “Pai-s/k: A robust automatic seismic p phase
arrival identification scheme,” IEEE Transactions on Geoscience and Remote Sensing, vol. 40, no. 6,
pp. 1395–1404, 2002.
[8] R. M. Allen and H. Kanamori, “The potential for earthquake early warning in southern california,”
American Association for the Advancement of Science, vol. 300, no. 5620, pp. 786–789, 2003.
[9] Y.-M. Wu, J.-K. Chung, T.-C. Shin, N.-C. Hsiao, Y.-B. Tsai, W. H. Lee, T.-l. Teng, et al., Development
of an integrated earthquake early warning system in Taiwan-Case for the Hualien area earthquakes.
National Emergency Training Center, 1999.
[10] Y.-M. Wu and L. Zhao, “Magnitude estimation using the first three seconds p-wave amplitude in
earthquake early warning,” Geophysical Research Letters, vol. 33, no. 16, 2006.
[11] Y.-M. Wu, “Progress on development of an earthquake early warning system using low-cost sensors,”
Pure and Applied Geophysics, vol. 172, no. 9, pp. 2343–2351, 2015.
[12] S. Megerian, F. Koushanfar, G. Qu, and M. Potkonjak, “Exposure in wireless ad-hoc sensor networks,”
in Proc. of MOBICOM, pp. 139–150, July 2001.
[13] B. Liu, O. Dousse, P. Nain, and D. Towsley, “Dynamic coverage of mobile sensor networks,” IEEE
Transactions on Parallel and Distributed systems, vol. 24, no. 2, pp. 301–311, 2013.
[14] H. V. Poor, An introduction to signal detection and estimation. Springer Science & Business Media,
2013.
[15] B. Krishnamachari and S. Iyengar, “Distributed bayesian algorithms for fault-tolerant event region
detection in wireless sensor networks,” IEEE Transactions on Computers, vol. 53, no. 3, pp. 241–250,
2004.
[16] T.-L. Chin and Y. H. Hu, “Optimal detector based on data fusion for wireless sensor networks,” in
IEEE Global Telecommunications Conference (GLOBECOM 2011), pp. 1–5, IEEE, 2011.
[17] J. N. Tsitsiklis, “Decentralized detection by a large number of sensors,” Mathematics of Control, Signals,
and Systems (MCSS), vol. 1, no. 2, pp. 167–182, 1988.
[18] T. Clouqueur, K. K. Saluja, and P. Ramanathan, “Fault tolerance in collaborative sensor networks for
target detection,” IEEE transactions on computers, vol. 53, no. 3, pp. 320–333, 2004.
[19] K. Yamanishi, J.-I. Takeuchi, G. Williams, and P. Milne, “On-line unsupervised outlier detection using
finite mixtures with discounting learning algorithms,” in Proceedings of the sixth ACM SIGKDD
international conference on Knowledge discovery and data mining, pp. 320–324, ACM, 2000.
[20] M. Davy, F. Desobry, A. Gretton, and C. Doncarli, “An online support vector machine for abnormal
events detection,” Signal processing, vol. 86, no. 8, pp. 2009–2025, 2006.
[21] S. Rajasegarar, C. Leckie, J. C. Bezdek, and M. Palaniswami, “Centered hyperspherical and hyperellipsoidal
one-class support vector machines for anomaly detection in sensor networks,” IEEE Transactions
on Information Forensics and Security, vol. 5, no. 3, pp. 518–533, 2010.
[22] C. Chiaruttini, V. Roberto, and F. Saitta, “Artificial intelligence techniques in seismic signal interpretation,”
Geophysical Journal International, vol. 98, no. 2, pp. 223–232, 1989.
[23] F. U. Dowla, S. R. Taylor, and R. W. Anderson, “Seismic discrimination with artificial neural networks:
preliminary results with regional spectral data,” Bulletin of the Seismological Society of America,
vol. 80, no. 5, pp. 1346–1373, 1990.
[24] M. Bevreuther, R. Carniel, and J. Wassermann, “Continuous hidden markov models: Application to
automatic earthquake detection and classification at las canadas caldera, tenerife,” Journal of volcanology
and geothermal research, vol. 176, no. 4, pp. 513–518, 2008.
[25] P. S. Dysart and J. J. Pulli, “Regional seismic event classification at the noress array: seismological
measurements and the use of trained neural networks,” Bulletin of the Seismological Society of
America, vol. 80, no. 6B, pp. 1910–1933, 1990.
[26] A. S. Alarifi, N. S. Alarifi, and S. Al-Humidan, “Earthquakes magnitude predication using artificial
neural network in northern red sea area,” Journal of King Saud University-Science, vol. 24, no. 4,
pp. 301–313, 2012.
[27] S. M. Mousavi, S. P. Horton, C. A. Langston, and B. Samei, “Seismic features and automatic discrimination
of deep and shallow induced-microearthquakes using neural network and logistic regression,”
Geophysical Journal International, vol. 207, no. 1, pp. 29–46, 2016.
[28] M. G. Vargas, J. Rueda, R. M. G. Blanco, and J. Mezcua, “A real-time discrimination system of earthquakes
and explosions for the mainland spanish seismic network,” Pure and Applied Geophysics,
vol. 174, no. 1, pp. 213–228, 2017.
[29] J. Wiszniowski, B. Plesiewicz, and J. Trojanowski, “Application of real time recurrent neural network
for detection of small natural earthquakes in poland,” Acta Geophysica, vol. 62, no. 3, pp. 469–485,
2014.
[30] M. Faulkner, M. Olson, R. Chandy, J. Krause, K. M. Chandy, and A. Krause, “The next big one: Detecting
earthquakes and other rare events from community-based sensors,” in Information Processing
in Sensor Networks (IPSN), 2011 10th International Conference on, pp. 13–24, IEEE, 2011.
[31] R.-C. Yin, Y.-M. Wu, and T.-Y. Hsu, “Application of the low-cost mems-type seismometer for structural
health monitoring: A pre-study,” in IEEE International Instrumentation and Measurement Technology
Conference (I2MTC), 2016.
[32] R. Allen, “Automatic earthquake recognition and timing from signal traces,” Bulletin of the seismological
Society of America, vol. 68, pp. 1521–1532, Oct. 1978.
[33] R. M. Allen and H. Kanamori, “The potential for earthquake early warning in southern california,”
American Association for the Advancement of Science, vol. 300, no. 5620, pp. 786–789, 2003.
[34] D. Coomans and D. L. Massart, “Alternative k-nearest neighbour rules in supervised pattern recognition:
Part 1. k-nearest neighbour classification by using alternative voting rules,” Analytica Chimica
Acta, vol. 136, pp. 15–27, 1982.
[35] W.-Y. Loh, “Classification and regression trees,” Wiley Interdisciplinary Reviews: Data Mining and
Knowledge Discovery, vol. 1, no. 1, pp. 14–23, 2011.
[36] R.-E. Fan, P.-H. Chen, and C.-J. Lin, “Working set selection using second order information for training
support vector machines,” Journal of machine learning research, vol. 6, pp. 1889–1918, Dec. 2005.