結合智慧感測標籤與機器學習方法判別混凝土內部鋼筋鏽蝕可能性之研究

NCUIR > Engineering College > Graduate Institute of Civil Engineering > Electronic Thesis & Dissertation > Item 987654321/91806

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/91806

Title:	結合智慧感測標籤與機器學習方法判別混凝土內部鋼筋鏽蝕可能性之研究
Authors:	黃柏勛;Huang, Bo-Xun
Contributors:	土木工程學系
Keywords:	鋼筋鏽蝕檢測;無損檢測;3D列印技術;RFID;ANN;steel reinforcement corrosion detection;non-destructive testing;3D printing technology;RFID;artificial neural network (ANN)
Date:	2023-07-25
Issue Date:	2024-09-19 14:14:27 (UTC+8)
Publisher:	國立中央大學
Abstract:	2019年發生南方澳大橋倒塌事件，是由於吊索系統鏽蝕而導致承載力不足，但鋼筋腐蝕通常發生於結構內部，難以從外部檢測。在該事件發生後，台灣開始強調了檢測鋼筋腐蝕對結構安全的重要性。而這事件也引導出一個問題，台灣社會與經濟的快速發展，都市人口密度上升，且因土地資源有限，無法隨意興建新的建築。為了有效利用土地資源，為結構進行補強，延長建築物壽命成為了一大命題。因此，需要一種快速且方便的檢測方法檢測結構狀況。本研究提出一種基於被動式無線感測技術和機器學習的新型結構檢測方式。利用低成本的感測器將其安裝在結構內部，並利用機器學習技術進行數據分析，來判斷結構中鋼筋是否存在腐蝕問題，實現低成本、快速和省人力的結構檢測。最終研究結果顯示，透過機器學習模型對過濾後的數據進行訓練，其對於鋼筋鏽蝕判別的正確率最高來到90.9 %，Recall也來到了93.3 %，表示對於鋼筋鏽蝕的判斷相當準確。該實驗同時表明，對有無鋼筋鏽蝕的梁進行破壞實驗，其裂縫表現也具有差異，而該差異便是能使梁在產生不同的吸水能力，然後依靠智慧感測標籤感測不同試體之間的水分差距，來完成判別鋼筋鏽蝕。本研究的結果有望提前發現鋼筋腐蝕的徵兆，並為結構補強提供依據，從而提高結構的安全性。相較於傳統的電化學檢測方法，本研究提出的新型檢測方式不僅成本更低，而且更具可行性和實用性。;The collapse of the Nanfang′ao Bridge in 2019 was caused by insufficient load-bearing capacity due to corrosion of the suspension system. However, steel reinforcement corrosion typically occurs within the structure and is difficult to detect externally. Following this incident, Taiwan began to emphasize the importance of detecting steel reinforcement corrosion for structural safety. This event also highlighted a challenge in Taiwan′s rapidly developing society and economy, where urban population density is increasing, and limited land resources restrict the construction of new buildings. In order to effectively utilize land resources and reinforce existing structures to extend their lifespan, a need arises for a fast and convenient method to assess structural conditions.Hence, there is a need for a fast and convenient inspection method for assessing structural conditions. This research proposes a novel structural inspection approach based on passive wireless sensing technology and machine learning. Low-cost sensors are installed inside the structure, and machine learning techniques are used for data analysis to determine the presence of corrosion in the steel reinforcement, achieving a low-cost, rapid, and labor-saving structural inspection process. The study′s findings indicate that the machine learning model, trained on filtered data, achieves a high accuracy rate of up to 90.9% for identifying steel reinforcement corrosion, with a recall rate of 93.3%, demonstrating accurate corrosion detection.The experiments also show that beams with and without steel reinforcement corrosion exhibit different crack patterns. This difference allows the beams to display varying water absorption abilities, and by using intelligent sensing tags to detect the moisture difference between different specimens, the identification of steel reinforcement corrosion can be accomplished.The results of this study are expected to detect early signs of steel reinforcement corrosion and provide a basis for structural reinforcement, thereby enhancing structural safety. Compared to traditional electrochemical testing methods, the proposed new inspection approach is not only more cost-effective but also more feasible and practical.
Appears in Collections:	[Graduate Institute of Civil Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	11	View/Open

社群 sharing

Loading...