Skip navigation

DSpace JSPUI

DSpace preserves and enables easy and open access to all types of digital content including text, images, moving images, mpegs and data sets

Learn More
DSpace logo
English
中文
  • Browse
    • Communities
      & Collections
    • Publication Year
    • Author
    • Title
    • Subject
    • Advisor
  • Search TDR
  • Rights Q&A
    • My Page
    • Receive email
      updates
    • Edit Profile
  1. NTU Theses and Dissertations Repository
  2. 工學院
  3. 應用力學研究所
Please use this identifier to cite or link to this item: http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/79366
Full metadata record
???org.dspace.app.webui.jsptag.ItemTag.dcfield???ValueLanguage
dc.contributor.advisor劉佩玲(Pei-Ling Liu)
dc.contributor.authorYuan-Tai Chenen
dc.contributor.author陳源泰zh_TW
dc.date.accessioned2022-11-23T08:59:00Z-
dc.date.available2021-11-03
dc.date.available2022-11-23T08:59:00Z-
dc.date.copyright2021-11-03
dc.date.issued2021
dc.date.submitted2021-10-27
dc.identifier.citation[1] Sansalone, M. and Carino, N. J., “Impact-Echo: A Method for Flaw Detection in Concrete Using Transient Stress Waves.” NBSIR 86-3452., Gaithersburg, MD:National Bureau of Standard 222., 1986. [2] Lin, Y., Sansalone, M. Carino, N. J., “Finite Element Studies of the Transient Response of Plates Containing Thin Layers and Voids.” J. Nondestructive Evaluation, vol. 9(1): pp. 27-47., 1990. [3] Lin, Y. and Sansalone, M., “Transient Response of Thick Circular and Square Bars Subjected to Transverse Elastic Impact.” J. Acoustical Society of Americ, vol. 91(2): pp. 885-893., 1992. [4] Lin, Y. and Sansalone, M., “Detecting Flaws in Concrete Beams and Columns Using the Impact-Echo Method.” Materials Journal of the American Concrete Institute, pp. 394-405., 1992. [5] Cheng, C. and Sansalone, M., “The Impact-Echo Response of Concrete Plates Containing Delaminations: Numerical, Experimental and Field Studies.” Material and Structures, vol. 26: pp. 274-285., 1992. [6] Wang, J.-J., et al., “Evaluation of Resonant Frequencies of Solid Circular Rods with Impact-Echo Method.” Journal of Nondestructive Evaluation, vol. 29(2): pp. 111-121., 2010. [7] Zein, A.S. and Gassman, S.L., “Frequency Spectrum Analysis of Impact-Echo Waveforms for T-Beams. Journal of Bridge Engineering”, vol. 15(6): pp. 705-714., 2010. [8] Hong, S.U., et al., “Estimation of slab depth, column size and rebar location of concrete specimen using impact echo method.” Materials Research Innovations, vol. 19: pp. 1167-1171., 2015. [9] Kee, S.-H. and Gucunski, N., “Interpretation of Flexural Vibration Modes from Impact-Echo Testing.” Journal of Infrastructure Systems, vol. 22(3), 2016. [10] Kohl, C. and Streicher, D., “Results of reconstructed and fused NDT-data measured in the laboratory and on-site at bridges.” Cement Concrete Composites, vol. 28(4): pp. 402-413., 2006. [11] Gucunski, N., et al., “Rapid Bridge Deck Condition Assessment Using Three-Dimensional Visualization of Impact Echo Data.” Journal of Infrastructure Systems, vol. 18(1): pp. 12-24., 2012. [12] Yao, F., et al., “Research on signal processing of segment-grout defect in tunnel based on impact-echo method.” Construction and Building Materials, vol. 187: pp. 280-289., 2018. [13] Yao, F. and Chen, G., “Time-Frequency Analysis of Impact Echo Signals of Grouting Defects in Tunnels.” Russian Journal of Nondestructive Testing, vol. 55(8): pp. 581-595., 2019. [14] Lin, Y. and Su, W.C., “The Use of Stress Waves for Determining the Depth of Surface-Opening Cracks in Concrete Structures.” ACI Materials Journal, vol. 93(5): pp. 494-505., 1996. [15] Liu, P.L., et al., “Scan of surface-opening cracks in reinforced concrete using transient elastic waves.” NDT E International, vol. 34(3): pp. 219-226., 2001. [16] Sun, Y., et al., “Depth estimation of surface-opening crack in concrete beams using impact-echo and non-contact video-based methods.” EURASIP Journal on Image and Video Processing, 2018(1): p. 144., 2018. [17] Wang, C.Y., et al., “Inspecting the current thickness of a refractory wall inside an operational blast furnace using the impact echo method.” Ndt E International, vol. 66: pp 43-51., 2014. [18] Xu, J.M., et al., “Damage detection of ballastless railway tracks by the impact-echo method.” Proceedings of the Institution of Civil Engineers-Transport, vol. 171(2): pp. 106-114., 2018. [19] LeCun, et al., “Deep learning. Nature”, vol. 521(7553): pp. 436-444., 2015. [20] Margrave, F., et al., “The use of neural networks in ultrasonic flaw detection.” Measurement, vol. 25(2): pp. 143-154., 1999. [21] Jarmulak, J., et al., “Case-based reasoning for interpretation of data from non-destructive testing.” Engineering Applications of Artificial Intelligence, vol. 14(4): pp. 401-417., 2001. [22] Hashimoto, K., et al., “Development of Autonomous Hammering Test Method for Deteriorated Concrete Structures Based on Artificial Intelligence and 3D Positioning System.” in Asset Intelligence through Integration and Interoperability and Contemporary Vibration Engineering Technologies., Springer. pp. 219-228., 2019. [23] Sarkar, S., et al. “Deep learning for structural health monitoring: A damage characterization application.” in Annual Conference of the Prognostics and Health Management Society. 2016. [24] Cha, Y.-J., et al., “Autonomous Structural Visual Inspection Using Region-Based Deep Learning for Detecting Multiple Damage Types.” Computer-Aided Civil and Infrastructure Engineering, vol. 33(9): pp. 731-747., 2018. [25] Beckman, G.H., et al., “Deep learning-based automatic volumetric damage quantification using depth camera.” Automation in Construction, vol. 99: pp. 114-124., 2019. [26] Sarmiento, J.S., et al., “Non-destructive Bridge Pavement Detection Using Impact Sound and Convolutional Neural Network.” in Proceedings of the 2019 5th International Conference on Computing and Artificial Intelligence. ACM., 2019. [27] Pratt, D. and Sansalone, M., “Impact-Echo Signal Interpretation Using Artificial Intelligence.” ACI Materials Journal, vol. 89(2): pp. 178-187., 1992. [28] Cohen L. “Time-Frequency Analysis. New Jersey: Prentice Hall”, 1995. [29] Shie Q. “Introduction to time-frequency and wavelet transforms.” Pearson Education Inc., 2002. [30] Yeh PL, Liu PL. “Application of the Wavelet Transform and the Enhanced Fourier Spectrum in the Impact Echo Test.” NDT E International. Vol. 41:pp. 382-94., 2008. [31] Yeh, P.L. and P.L. Liu, “Application of the Wavelet Transform and the Enhanced Fourier Spectrum in the Impact Echo Test.” NDT E International, vol. 41(5): pp. 382-394., 2008. [32] Jeong, J. and Williams, W.J., “Kernel design for reduced interference distributions”, 1992. [33] Boukaye BoubacarTraore, et al., “Deep convolution neural network for image recognition.”, 2018 [34] Alexey Bochkovskiy, et al., “YOLOv4: Optimal Speed and Accuracy of Object Detection.”, 2020. [35] Goldsmith, W. “Impact: The Theory and Physical Behavior of Colliding Solids.” London: Edward Arnold Ltd., 1965. [36] Colla and Lausch, “Influence of Source Frequency on Impact-echo Data Quality for Testing Concrete Structure”, NDT E International, Vol. 36, pp203.-213., 2003.
dc.identifier.urihttp://tdr.lib.ntu.edu.tw/jspui/handle/123456789/79366-
dc.description.abstract敲擊回音法為一種廣泛應用於檢測混凝土結構的非破壞檢測技術。以敲擊回音法量測之時間域訊號藉由傅立葉轉換可轉換為頻率域訊號,具有較高能量的頻率分量會在傅立葉頻譜中形成尖峰,檢測員可以據此來檢測混凝土中裂縫的位置。 然而,不僅來自裂縫的回波會在頻譜中形成尖峰,模態振動也是如此。表面波也在頻譜中也會形成一個駝峰。振動之尖峰和表面波之尖峰的存在有時會妨害時頻圖中裂縫回波的檢測,因為與振動和表面波相比,裂縫回波通常較弱。因此,本研究使用時頻分析將時間訊號轉換為時頻圖。由於回波、表面波和模態振動的持續時間不同,我們可以通過觀察時頻圖中亮帶長度的差異來更好地區分它們。 過去,時頻圖的解釋依賴於經驗豐富的檢測員手動完成。這是耗時且不可靠的,因為人類會犯錯。本研究的目標是開發一種基於深度學習方法的識別系統,可以在時頻圖中識別回波、表面波和振動,並自動計算裂縫的深度(如果有的話)。 卷積神經網路(CNN)為一種強大的深度學習方法。可以訓練它對圖像中出現的對象進行分類。 You only look once (YOLO) 為另一種深度學習方法,它可以在單個圖像中進行多個對象檢測。本研究同時使用了 CNN 迴歸模型和 YOLO V4 模型來判讀敲擊回音時頻圖。 CNN迴歸模型的輸入為時頻圖,輸出為裂縫深度。 YOLO V4 模型的輸入也是時頻圖,輸出則是帶註釋的邊界框,包括“裂縫”框、“表面波”框和“振動”框,以及裂縫的深度,其中裂縫深度是使用“裂縫”框之位置進行計算。 CNN 回歸模型是使用給定裂縫深度的時頻圖進行訓練。 YOLO V4 模型則是使用預先標註之時頻圖進行訓練,該圖像已預先標註了裂縫、表面波和振動,並給予裂縫深度。 在訓練模型時,考慮了幾個問題,包括色階、正規化方法、時頻分析方法、雜訊的加入、圖像解析度、訓練數據集的組成以及混凝土試體的邊界條件。產生最佳結果的模型如下: 1. 使用 YOLO V4 模型, 2. 使用灰階 RID 時頻圖作為輸入, 3. 時頻圖自動正規化, 4. 圖像的解析度為 , 5. 訓練集包括含有雜訊之數據和實驗數據; 6.訓練集中包含半無限空間模擬數據。對於模擬數據,裂縫檢測率為99.7%,深度預測誤差為2.3%。對於實驗數據,裂縫檢測率為95.6%,深度預測誤差為2.8%。 最佳模型的裂縫檢測率和深度預測誤差是令人滿意的。期望藉助該模型,可以大大地減少檢測誤差和檢測時間,並同時提高判讀結果的準確性。zh_TW
dc.description.provenanceMade available in DSpace on 2022-11-23T08:59:00Z (GMT). No. of bitstreams: 1
U0001-2510202120105700.pdf: 4430367 bytes, checksum: 0f33de3fb9a5c9e0aade9816d45f3df2 (MD5)
Previous issue date: 2021
en
dc.description.tableofcontents目錄 致謝 I 摘要 II Abstract IV 目錄 VI 圖目錄 IX 表目錄 XIII 第一章 前言 1 1.1 研究動機 1 1.2 文獻回顧 2 1.3 全文簡介 4 第二章 敲擊回音法 5 2.1 敲擊回音法之原理 5 2.2 敲擊源 8 2.3 時頻分析方法 9 2.3.1 Wignner-Ville Distribution 10 2.3.2 Reduce Interference Distribution 12 第三章 深度學習模型 14 3.1 CNN迴歸模型 14 3.1.1 卷積層 15 3.1.2 池化層 16 3.1.3 激活函數(Active Function) 17 3.1.4 損失函數(Loss Function) 17 3.1.5 最佳化(Optimization) 18 3.2 YOLO V4模型 19 3.2.1 YOLO V4模型架構 19 3.2.2 模型評估指標 21 第四章 資料集 24 4.1 數值模擬資料 24 4.1.1 數值模擬敲擊回音試驗步驟 25 4.1.2 數值模擬試體介紹 28 4.1.3 數值模擬訊號產生雜訊 30 4.2 實驗資料 32 4.2.1 實驗設備 32 4.2.2 實驗訊號測量與接收 34 4.2.3 實驗試體介紹 34 4.2.4 時間域訊號原點調整 35 4.3 訓練集與測試集 38 第五章 結果與討論 43 5.1 CNN迴歸模型測試結果 43 5.1.1 色階影響 43 5.1.2 雜訊影響 45 5.1.3 正規化方法比較 48 5.2 YOLO V4模型測試結果 51 5.2.1 時頻分析方法比較 53 5.2.2 雜訊影響 55 5.2.3 解析度影響 57 5.2.4 加入實驗數據訓練之影響 60 5.2.5 底部邊界比較 62 5.2.6 YOLO V4模型泛用性測試 67 5.2.7 表面波與模態辨識結果 68 5.2.8 資料集輪替測試結果 70 5.3 YOLO V4模型參數調整 73 5.3.1 Max batches 73 5.3.2 Ignore threshold 74 第六章 結論與未來展望 75 參考文獻 77
dc.language.isozh-TW
dc.title以深度學習方法判讀敲擊回音時頻圖zh_TW
dc.titleThe Interpretation of Impact-Echo Spectrogram Using Deep Learning Methodsen
dc.date.schoolyear109-2
dc.description.degree碩士
dc.contributor.oralexamcommittee林宜清(Hsin-Tsai Liu),孫嘉宏(Chih-Yang Tseng)
dc.subject.keyword敲擊回音法,時頻分析,非破壞檢測,深度學習,卷積神經網路,YOLO V4,zh_TW
dc.subject.keywordImpact-Echo Method,Time-Frequency Analysis,Nondestructive Testing,Deep Learning,CNN,YOLO V4,en
dc.relation.page80
dc.identifier.doi10.6342/NTU202104177
dc.rights.note同意授權(全球公開)
dc.date.accepted2021-10-28
dc.contributor.author-college工學院zh_TW
dc.contributor.author-dept應用力學研究所zh_TW
Appears in Collections:應用力學研究所

Files in This Item:
File SizeFormat 
U0001-2510202120105700.pdf4.33 MBAdobe PDFView/Open
Show simple item record


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

社群連結
聯絡資訊
10617臺北市大安區羅斯福路四段1號
No.1 Sec.4, Roosevelt Rd., Taipei, Taiwan, R.O.C. 106
Tel: (02)33662353
Email: ntuetds@ntu.edu.tw
意見箱
相關連結
館藏目錄
國內圖書館整合查詢 MetaCat
臺大學術典藏 NTU Scholars
臺大圖書館數位典藏館
本站聲明
© NTU Library All Rights Reserved