台灣位於板塊交界處，地震頻繁發生，人民的生命財產安全受到
危害，房屋的耐震能力面臨巨大挑戰。為應對地震所帶來的風險，本
研究提出一結合深度學習與房屋街景圖像之機率式地震風險評估方
法，針對地震高風險區域建築之耐震能力作進一步的分析與探討，減
少地震所帶來的危害。以傳統方法進行建築物耐震分析，成本較高且
相當費時，倘若要進行大範圍風險評估此法不可行。因此本研究嘗試
透過深度學習獲取建築物高度，藉由地震工程模擬軟體建立數值模
型，模擬結構之受震情形。本研究採用機率式建物損害耐震評估架構
進行非線性動力歷時分析，損害判定準則參考 FEMA 技術報告之
Hazus 4.2-SP3 RC 結構損害等級，統計回歸成損害等級易損曲線後計
算建築物損害機率，以此方法量化建築物受震時所造成的損害程度。
最後本研究繪製花蓮縣之地震風險評估地圖，與 0403 花蓮大地震之
實際災情比較與驗證該方法之可行性，期望可做為日後機率式地震風
險評估之建立標準流程與範本。

Taiwan is located at the intersection of tectonic plates, resulting in
frequent earthquakes that threaten the safety of people's lives and property.
The seismic resilience of buildings faces significant challenges. To address
the risks posed by earthquakes, this study proposes a probabilistic seismic
risk assessment method that combines deep learning with street view
images of buildings. This method aims to further analyze and evaluate the
seismic resilience of buildings in high-risk earthquake areas, thereby
mitigating the damage caused by earthquakes.Traditional methods for
seismic analysis of buildings are costly and time-consuming, making them
impractical for large-scale risk assessments. Therefore, this study attempts
to use deep learning to obtain building heights and employs seismic
engineering simulation software to create numerical models that simulate
the structural response to earthquakes.The study adopts a probabilistic
seismic damage assessment framework to perform nonlinear dynamic
time-history analysis. The damage assessment criteria refer to the Hazus
4.2-SP3 RC structure damage levels from the FEMA technical report. By
statistically regressing the damage levels into fragility curves, the study
calculates the probability of building damage. This method quantifies the
extent of damage caused by earthquakes to buildings.Finally, this study
maps the seismic risk assessment of Hualien County and compares it with
the actual damage from the April 3 Hualien earthquake to verify the
feasibility of the proposed method. It is hoped that this can serve as a
standard procedure and template for probabilistic seismic risk assessment
in the future.

[1] 葉勇凱, & 周德光. (2019). 土木404-100設計例單自由度簡化模型
之易損曲線. 國家地震工程研究中心, NCREE-19-023.
[2] 林瑞良, 劉子豪, & 蔡克銓. (2024). 以簡化方法分析具非比例阻
尼之平面不對稱結構. 國家地震工程研究中心, NCREE-14-001.
[3] Vijay badrinarayanan, Ankur handa, & Roberto cipolla. (2015). SegNet: A Deep
Convolutional Encoder-Decoder Architecture for Robust Semantic Pixel-Wise
Labelling.ArXiv,1505.07293.
[4] Olaf ronneberger, Philipp fischer, & Thomas brox. (2015). U-Net: Convolutional
Networks for Biomedical Image Segmentation.Medical Image Computing and
Computer-Assisted Intervention–MICCAI 2015: 18th International Conference,
Munich,Germany,October5-9,2015,Proceedings,PartIII18, 234–241.
[5] Bulent siyah. (2021).DeepLearningBasedSemanticSegmentation|Keras. Kaggle.
https://www.kaggle.com/code/bulentsiyah/deep-learning-based-semanticsegmentation-keras
[6] Sambit pattanaik, Agbotiname lucky imoize, Chun-ta li, Sharmila anand john francis,
Cheng-chi lee, & Diptendu sinha roy. (2023). Data-Driven Diffraction Loss
Estimation for Future Intelligent Transportation Systems in 6G
Networks.Mathematics.
[7] Ala’a al-habashna. (2021).Building Height Estimation Using Street-View Images,
Deep-Learning,ContourProcessing,andGeospatialData(2021 18th Conference on
Robots and Vision (CRV)). IEEE.
[8] Xiaojiang li, Chuanrong zhang, & Weidong li. (2021). Does the Visibility of Greenery
Increase Perceived Safety in Urban Areas? Evidence from the Place Pulse 1.0
Dataset.ISPRSInternationalJournalofGeo-Information,ISSN2220-9964.
[9] Yann lecun, Léon bottou, Yoshua bengio, & Patrick haffner. (1998). Gradient-Based
Learning Applied to Document Recognition.ProceedingsoftheIEEE,86(11).
[10] Alex krizhevsky, Ilya sutskever, & Geoffrey e hinton. (2012). 2012 AlexNet.Adv.
NeuralInf.Process.Syst, 1–9.
[11] Karen simonyan, & andrew zisserman. (2015). Very Deep Convolutional Networks
for Large-Scale Image Recognition.ArXiv,1409.1556.
[12] Shaoqing ren, kaiming he, Xiangyu zhang, & Jian sun. (2016). Deep Residual
Learning for Image Recognition.IEEE Computer Vision and Pattern Recognition
(CVPR),pp. 770–778.
[13] Dawei du, Pengfei zhu, & Longyin wen. (2019). VisDrone-DET2019: The Vision
Meets Drone Object Detection in Image Challenge Results.Proceedings of the
IEEE/CVFInternationalConferenceonComputerVisionWorkshops,pp. 0–0.
[14] Sal saad al deen taher, ji dang. (2022). Autonomous Multiple Damage Detection and
Segmentation in Structures Using Mask R-CNN.Experimental VibrationAnalysis
forCivilEngineeringStructures:SelectProceedingsoftheEVACES2021, 545–556.
[15] Zhixuhao. (2019).Implementation of Deep Learning Framework -- Unet, Using Keras. Github. https://github.com/zhixuhao/unet
[16] Jyun-cheng li. (2023).AStudyonCorrosion IdentificationofSteel StructureBridge
withDeepLearningModelMaskR-CNN.
[17] Guo-bin zhou. (2021).SeismicRecognitionofSmall-to-LargeSizedLandslidewith
Machine-andDeep-LearningMethod.
[18] Zongwei zhou, Md mahfuzur rahman siddiquee, Nima tajbakhsh, & Jianming liang.
(2018). UNet++: A Nested U-Net Architecture for Medical Image
Segmentation.DeepLearninginMedicalImageAnalysisandMultimodalLearning
forClinicalDecisionSupport, 3–11.
[19] Huimin huang, lanfen lin, ruofeng tong, hongjie hu, qiaowei zhang, yutaro iwamoto,
Xianhua han, Yen-wei chen, & jian wu. (2020). UNet 3+: A Full-Scale Connected
UNet for Medical Image Segmentation.ArXiv,2004.08790.
[20] Jesline daniel, J. t. anita rose, F. sangeetha francelin vinnarasi, & Venkatesan
rajinikanth. (2022). VGG-UNet/VGG-SegNet Supported Automatic Segmentation
of Endoplasmic Reticulum Network in Fluorescence Microscopy Images.Scanning
2022,e9243, 1–11.
[21] ACT-40, Seismic evaluation and retrofit of concrete buildings. Report No. SSC 96-
01, Applied Technology Council, 1996.
[22] Porter, K. A. (2003, July). An overview of PEER’s performance-based earthquake
engineering methodology. InProceedings of ninth international conference on
applications of statistics and probability in civil engineering(pp. 1-8).
[23] Moehle, J., & Deierlein, G. G. (2004, August). A framework methodology for
performancebased earthquake engineering. In13th world conference on earthquake
engineering(Vol. 679, p. 12). WCEE Vancouver.
[24] FEMA P-58. (2012). Federal Emergency Management Agency: Seismic
Performance Assessment of Buildings.
[25] Vicente, R., Ferreira, T., & Maio, R. (2014). Seismic risk at theurban scale:
assessment, mapping and planning. Procedia Economics and Finance, 18, 71-80.
[26] Pelà, L. (2018). New trends and challenges in large-scale and urban assessment of
seismic risk in historical centres. International Journal of Architectural Heritage, 12(7-
8), 1051-1054
[27] Ahmed, S., Abarca, A., Perrone, D., & Monteiro, R. (2022). Large-scale seismic
assessment of RC buildings through rapid visual screening. International Journal of
Disaster Risk Reduction, 80, 103219.
[28] Roberto cipolla, Alex kendall, & Vijay badrinarayanan. (2017). Segnet: A Deep
Convolutional Encoder-Decoder Architecture for Image Segmentation.IEEE
TransactionsonPatternAnalysisandMachineIntelligence,39(12), 2481–2495.
[29] Jonathan long, Evan shelhamer, & Trevor darrellnan. (2015). Fully Convolutional
Networks for Semantic Segmentation.ArXiv,1411,4038.
[30] Olaf ronneberger, Thorsten falk, & Dominic mai. (2019). U-Net: Deep Learning for
Cell Counting, Detection, and Morphometry.NatureMethods,16(1), 67–70.
[31] Chuanqi tan, fuchun sun, tao kong, wenchang zhang, chao yang, & Chunfang liu.(2018). A Survey on Deep Transfer Learning.ArXiv,1808.01974.
[32] Kentaro wada. (2019).Labelme. Github. https://github.com/labelmeai/labelme
[33] Anindya apriliyanti pravitasari , Nur iriawan, Mawanda almuhayar, & Taufik azmi.
(2020). UNet-VGG16 with Transfer Learning for MRI-Based Brain Tumor
Segmentation.TELKOMNIKA,18(3), 1310.
[34] Sepp hochreiterti pravitasari , & Jurgen schmidhuber. (1997). LONG SHORTTERM MEMORY.IDSIA,9(8), 1735–1780.
[35] Yann lecun, & Yoshua bengio. (1995). Convolutional Networks for Images, Speech,
and Time Series.The Handbook of BrainTheory and Neural Networks,3361(10),1995.
[36] D. E. Rumelhart, G. E. Hinton, and R. J. Williams, "Learning representations by backpropagating errors," Nature, vol. 323, no. 6088, pp. 533-536, 1986/10/01 1986.
[37] Jiahui yu, Yuning jiang, Zhangyang wang, Zhimin cao, & Thomas huang. (2016).
UnitBox: An Advanced Object Detection Network.ArXiv,1608.01471.
[38] Rezania azdy, & Febriyanti darnis. (2022). Use of Haversine Formula in Finding
Distance Between Temporary Shelter and Waste End Processing Sites.Journal of
Physics:ConferenceSeries,1500(1), 012104.
[39] 國 家 地 震 工 程 研 究 中 心 . (2018). 建 物 勘 災 資 料 庫 .
https://www.ncree.org/recce/search.aspx?eq=20180206HLTW
[40] Richard hogan. (2015).How to Calculate Normalized Error. Isobudgets.
https://www.isobudgets.com/how-to-calculate-normalized-error/
[41] 李坤展. (2023). 數據驅動之鋼筋混凝土構架機率式地震風險評估.碩士論文，國立中央大學土木工程學系(2023).
[42] Nathan m. newmark. (1959).A Method of Computation for Structural
Dynamics.85(EM3),pp. 67–94.
[43] Amirahmad fathiehrk. (2013).Nonlinear Dynamic Analysis of Modular Steel
BuildingsinTwoandThreeDimensions.
[44] Ibarra L.F., Medina R. A., and Krawinkler H. (2005). “Hysteretic models that
incorporate strength and stiffness deterioration”, Earthquake Engineering and
Structural Dynamics, 34(12), 1489-1511.
[45] Karavasilis T.L., Ricles J.M., Sause R. (2009). "Implementation of deterioration
elements in OpenSEES for collapse simulations" ATLSS Engineering Research
Center, Rep. No. 09-11.
[46] Lignos, D.G., and Krawinkler, H. (2011). “Deterioration modeling of steel
components in support of collapse prediction of steel moment frames under earthquake loading”, Journal of Structural Engineering, ASCE, Vol. 137 (11), 1291-1302.
[47] Lignos, D.G., Krawinkler, H. (2012). “Development and Utilization of Structural
Component Databases for Performance-Based Earthquake Engineering", Journal of
Structural Engineering, ASCE, doi: 10.1061/(ASCE)ST.1943-541X.0000646
[48] Mazzoni, S., McKenna, F., Scott, M. H., & Fenves, G. L. (2006). OpenSees
command language manual.Pacific Earthquake Engineering Research (PEER) Center,264(1), 137-158.
[49] 內政部營建署(2022)。建築物耐震設計規範及解說，中華民國內政部營建署，
台北，台灣。
[50] FEMA P-58. (2012). Federal Emergency Management Agency: Seismic
Performance Assessment of Buildings.
[51] 劉勛仁 , 簡文郁, & 張毓文. (2018). 台灣泛域工址設計用實測地震歷時篩選
研究.中華民國第 15 屆結構工程及第 5 屆地震工程研討會,143.
[52] Fema. (2020).HazusEarthquakeModelTechnicalManual.Hazus4.2SP3.
[53] Chen, P. Y., & Guan, X. (2023).A multi-source data-driven approach for evaluating
the seismic response of non-ductile reinforced concrete moment frames. Engineering
Structures, 278, 115452.