| 研究生: |
黃冠淯 Huang, Guan-Yu |
|---|---|
| 論文名稱: |
應用被動式熱成像與機器學習於外牆瓷磚空鼓缺陷檢測之研究 Detection of Delamination in Exterior Wall Tiles via Passive Thermography and Machine Learning |
| 指導教授: |
馮重偉
Feng, Chung-Wei |
| 學位類別: |
碩士 Master |
| 系所名稱: |
工學院 - 土木工程學系 Department of Civil Engineering |
| 論文出版年: | 2025 |
| 畢業學年度: | 113 |
| 語文別: | 中文 |
| 論文頁數: | 134 |
| 中文關鍵詞: | 被動式熱成像 、機器學習 、外牆瓷磚空鼓缺陷 、非破壞性檢測 |
| 外文關鍵詞: | Passive Thermography, Machine Learning, Exterior tile delamination, Non-destructive testing (NDT) |
| 相關次數: | 點閱:6 下載:0 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
近年來,高樓建築外牆瓷磚剝落事件頻傳,嚴重威脅公共安全。傳統外牆檢測方法,如搭設鷹架或人員垂降,不僅成本高昂、耗時且伴隨作業風險,其檢測範圍與結果判讀亦高度依賴人工經驗,導致全面性的定期檢測難以有效落實。面對大量檢測需求與數據,發展可靠、高效率且具成本效益的自動化檢測技術,尤以整合人工智慧與大數據分析者為當務之急。
本研究旨在提出整合被動式熱成像與機器學習的外牆瓷磚空鼓缺陷自動檢測方案。首先以標準化打診法標註空鼓狀態,作為熱成像模型訓練目標。繼而採集牆面被動式熱影像,提取溫度變異特徵。基於此數據集與打診標註,訓練評估支持向量機(SVM)、隨機森林(RF)及邏輯回歸(LR)三種機器學習模型,以建構能從大量熱影像中自動辨識空鼓的智慧系統。
結果顯示,於訓練數據上,隨機森林模型表現最佳,其宏平均F1分數達到0.99,支持向量機則次之為0.96,初步驗證了本方案之可行性。然而,在後續的獨立案例驗證中,模型面對全新數據時的泛化能力仍有待提升,此結果明確指出了未來研究的優化方向。儘管如此,本研究成功建立了一套完整的自動化檢測框架,證明其在提供一個具備大數據處理能力且更經濟的替代方案上具有重要價值,可望提升檢測頻率與建築物公共安全。
The frequent peeling of exterior wall tiles from high-rise buildings has become a serious threat to the public. Traditional inspection methods, such as scaffolding and rope access, are costly, time-consuming, and hazardous. Furthermore, their reliance on manual experience limits the scope and reliability of assessments. This highlights an urgent need for reliable and efficient automated technologies that integrate artificial intelligence and big data analytics.
This study proposes an automated solution for detecting debonding defects in exterior tiles by combining passive thermography with machine learning. Initially, a standardized tapping method was used to label the debonding status of tiles, serving as ground truth for model training. Passive thermal images of the wall surfaces were then captured to extract features based on temperature variations. Using this dataset, three machine learning models—Support Vector Machine (SVM), Random Forest (RF), and Logistic Regression (LR)—were trained and evaluated to build an intelligent system for automated defect identification.
Results demonstrate the Random Forest model achieved the best performance on the training data, with a macro-average F1-score of 0.99, followed by the SVM at 0.96, preliminarily validating the approach's feasibility. However, in subsequent independent case validations, the models' generalization capability on new data indicated a need for further improvement, clearly defining future research directions. Nevertheless, this study successfully establishes a complete automated inspection framework, proving its value as a more economical alternative with big data processing capabilities that can enhance inspection frequency and public safety.
[1] 高雄市政府工務局建築管理處(2015)。高雄市政府103年度6層樓以上建築物之外牆安全檢查執行成果。高雄市政府全球資訊網。2025年05月查詢,取自 https://build.kcg.gov.tw/index.aspx?au_id=12&sub_id=35&id=163699
[2] 廖硃岑、張孟修、林智捷(2022)。外牆非破壞與破壞檢測方法之整合研究。物業管理學報,13(2),39-52。2025年07月查詢,取自 https://www.airitilibrary.com/Article/Detail/20765509-202209-202210260010-202210260010-39-52
[3] 厲娓娓(2017)。建築物外牆瓷磚飾材評估檢查及診斷標準之研究(內政部建築研究所自行研究報告)。內政部建築研究所。2025年07月查詢,取自https://www.abri.gov.tw/News_Content_Table.aspx?n=807&s=39511
[4] 臺北市建築管理工程處(2024a)。臺北市建築物外牆安全診斷檢查及申報辦法。臺北市法規查詢系統。2025年05月查詢,取自https://www.laws.taipei.gov.tw/Law/LawSearch/LawArticleContent/FL092871
[5] 臺北市建築管理工程處(2024b)。臺北市建築物外牆安全診斷檢查及申報相關費用統計清冊。臺北市政府網站。2025年05月查詢,取自https://cmo.gov.taipei/cp.aspx?n=417EB4020336EFEB
[6] 臺北市建築管理工程處(2024c)。附表一臺北市建築物外牆安全檢查之檢查報告書。臺北市政府網站。2025年05月查詢,取自https://cmo.gov.taipei/cp.aspx?n=417EB4020336EFEB
[7] 內政部不動產資訊平台(2025)。統計資訊主題查詢。內政部不動產資訊平台網站。2025年01月查詢,取自https://pip.moi.gov.tw/Publicize/Info/E4010
[8] 臺北市建築管理工程處(2024d)。專業檢查人員認可證申請書。臺北市政府網站。2025年05月查詢,取自https://cmo.gov.taipei/cp.aspx?n=417EB4020336EFEB
[9] 翁嘉南(2017)。以超音波方法量測磁磚黏著層介面之研究。﹝碩士論文。國立高雄應用科技大學﹞臺灣博碩士論文知識加值系統。取自https://hdl.handle.net/11296/2vp9ay。
[10] 戴佩宜(2008)。以打音法從事建築外牆瓷磚非破壞檢測之研究。﹝碩士論文。國立高雄大學﹞臺灣博碩士論文知識加值系統。取自https://hdl.handle.net/11296/6jqun4。
[11] 許家豪(2010)。建築物外牆磁磚與木構件內部缺陷檢測之研究。﹝碩士論文。國立成功大學﹞臺灣博碩士論文知識加值系統。取自https://hdl.handle.net/11296/v34645。
[12] 洪子傑(2016)。利用打音法及頻譜分析診斷建築物外牆磁磚劣化之研究〔碩士論文,國立臺灣大學〕。華藝線上圖書館。取自https://doi.org/10.6342/NTU201600543
[13] 林谷陶(2011)。紅外線熱影像法於外牆磁磚表面溫度檢測特性之實驗研究(內政部建築研究所自行研究報告)。內政部建築研究所。2025年07月查詢,取自https://www.abri.gov.tw/News_Content_Table.aspx?n=807&s=38873
[14] 朱瑞鼎和厲學成(2013)。探究臺南地區的氣溫特性及其影響。(102 年度研究報告第 CWB102-1A-15 號)。臺灣南區氣象中心。2025年07月查詢,取自https://south.cwa.gov.tw/fileApply/category/Rrqr1565227059ASAz?page=3
[15] 湯永州(2017)。紅外線熱影像技術應用於外牆磁磚檢測與風速關聯性之研究。﹝碩士論文。國立高雄大學﹞臺灣博碩士論文知識加值系統。取自https://hdl.handle.net/11296/b93et9。
[16] 江立偉、羅紫萍(2015)。臺灣老舊建築物外牆磁磚更新評估模式分析之研究。載於台灣物業管理學會(主編),物業管理學會論文集(頁283-289)。台灣物業管理學會。取自https://www.airitilibrary.com/Article/Detail?DocID=P20130903002-201507-201509170005-201509170005-283-289
[17] 陳瑞鈴、張奇偉、林鎮華、張竣堯、周楷峻、厲娓娓、黃智偉、程冠銘(2017)。建築外牆瓷磚非破壞性檢測之研究-以紅外線熱顯像技術結合無人飛行載具為例(內政部建築研究所協同研究報告)。內政部建築研究所。2025年07月查詢,取自https://www.abri.gov.tw/News_Content_Table.aspx?n=807&s=39518
[18] Tan, K. S., Chan, K. C., Wong, B. S., & Guan, L. W. (1996). Ultrasonic evaluation of cement adhesion in wall tiles. Cement and Concrete Composites, 18(2), 119–124. Retrieved from https://doi.org/10.1016/0958-9465(95)00008-9
[19] Hussain Alsalem, a., Tanakab, T., Hondac, T., Doi, S., & Uchidae, S. (2020). Measuring Adhesion Strength of Wall Tile to Concrete by Non-Contact Inspection Using Electromagnetic Waves. In Proceedings of the 37th International Symposium on Automation and Robotics in Construction (ISARC) (pp. 633-638). Kitakyushu, Japan: IAARC. Retrieved from https://doi.org/10.22260/ISARC2020/0086
[20] Tomita, K., & Chew, M. Y. L. (2022). A Review of Infrared Thermography for Delamination Detection on Infrastructures and Buildings. Sensors, 22(2), 423. Retrieved from https://doi.org/10.3390/s22020423
[21] Washer, G., Fenwick, R., & Bolleni, N. (2010). Effects of solar loading on infrared imaging of subsurface features in concrete. Journal of Bridge Engineering, 15(4), 384–390. Retrieved from https://doi.org/10.1061/(ASCE)BE.1943-5592.0000117
[22] Bauer, E., Pavón, E., Barreira, E., & De Castro, E. K. (2016). Analysis of building facade defects using infrared thermography: Laboratory studies. Journal of Building Engineering, 6, 93–104. Retrieved from https://doi.org/10.1016/j.jobe.2016.02.012
[23] Chiang, L.-W., & Guo, S.-J. (2017). The Research of Infrared Thermal Image Diagnostic Model for Building External Wall Tiles. Advances in Technology Innovation, 2(2), 56–60. Retrieved from https://ojs.imeti.org/index.php/AITI/article/view/277
[24] Behbahani, S. S., Golrokh, A. J., Hafiz, A., Wiggenhauser, H., & Tabrizi, K. (2024). Utilizing infrared thermography for the condition assessment of tunnel lining with tiled surface in various temperature conditions. Tunnelling and Underground Space Technology, 154, 106093. Retrieved from https://doi.org/10.1016/j.tust.2024.106093
[25] Maldague, X., & Marinetti, S. (1996). Pulse phase infrared thermography. Journal of Applied Physics, 79(5), 2694–2698. Retrieved from https://doi.org/10.1063/1.362662
[26] Rajic, N. (2002). Principal component thermography for flaw contrast enhancement and flaw depth characterisation in composite structures. Composite Structures, 58(4), 521–528. Retrieved from https://doi.org/10.1016/S0263-8223(02)00161-7
[27] Shepard, S. M., Lhota, J. R., Rubadeux, B. A., Wang, D., & Ahmed, T. (2003). Reconstruction and Enhancement of Active Thermographic Image Sequences. Optical Engineering, 42(5), 1337-1342. Retrieved from https://doi.org/10.1117/1.1566969
[28] Salcedo-Sanz, S., Rojo-Álvarez, J.L., Martínez-Ramón, M. and Camps-Valls, G. (2014), Support vector machines in engineering: an overview. WIREs Data Mining Knowl Discov, 4: 234-267. Retrieved from https://doi.org/10.1002/widm.1125
[29] Random Forest Algorithm Overview (H. A. Salman, A. Kalakech, & A. Steiti , Trans.). (2024). Babylonian Journal of Machine Learning, 2024, 69-79. Retrieved from https://doi.org/10.58496/BJML/2024/007
[30] Healy, L. M. (2006) ‘Logistic regression: An overview’, Eastern Michighan College of Technology. Retrieved from https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=bb110f17a7b5127aec8c1e302e40b2b3c9ef18a7
[31] Tong, Z., Cheng, L., Xie, S., & Kersemans, M. (2023). A flexible deep learning framework for thermographic inspection of composites. NDT & E International, 139, 102881. Retrieved from https://doi.org/10.1016/j.ndteint.2023.102881