• Smart Structures and Systems
• /
• 제31권4호
• /
• pp.365-381
• /
• 2023

The existing vision-based techniques for inspection and condition assessment of civil infrastructure are mostly manual and consequently time-consuming, expensive, subjective, and risky. As a viable alternative, researchers in the past resorted to deep learning-based autonomous damage detection algorithms for expedited post-disaster reconnaissance of structures. Although a number of automatic damage detection algorithms have been proposed, the scarcity of labeled training data remains a major concern. To address this issue, this study proposed a semi-supervised learning (SSL) framework based on consistency regularization and cross-supervision. Image data from post-earthquake reconnaissance, that contains cracks, spalling, and exposed rebars are used to evaluate the proposed solution. Experiments are carried out under different data partition protocols, and it is shown that the proposed SSL method can make use of unlabeled images to enhance the segmentation performance when limited amount of ground truth labels are provided. This study also proposes DeepLab-AASPP and modified versions of U-Net++ based on channel-wise attention mechanism to better segment the components and damage areas from images of reinforced concrete buildings. The channel-wise attention mechanism can effectively improve the performance of the network by dynamically scaling the feature maps so that the networks can focus on more informative feature maps in the concatenation layer. The proposed DeepLab-AASPP achieves the best performance on component segmentation and damage state segmentation tasks with mIoU scores of 0.9850 and 0.7032, respectively. For crack, spalling, and rebar segmentation tasks, modified U-Net++ obtains the best performance with Igou scores (excluding the background pixels) of 0.5449, 0.9375, and 0.5018, respectively. The proposed architectures win the second place in IC-SHM2021 competition in all five tasks of Project 2.

• 콘크리트학회논문집
• /
• 제22권5호
• /
• pp.719-725
• /
• 2010

최근 콘크리트 구조물에 충격하중, 폭발하중 등 극한의 외력이 작용하는 경우가 빈번하게 발생하고 있다. 이 연구에서는 강섬유보강 RC보와 carbon FRP 시트를 이용하여 보강한 RC보를 이용하여 충격실험을 진행하였다. 강섬유보강 RC보의 경우 0.75%부피비로 강섬유를 혼입하였으며 carbon FRP 시트의 경우 에폭시 레진을 이용하여 보강을 한후 보 부재를 완성하였다. FRP 시트 보강은 부재의 하단을 휨 보강하였으며 충격하중이 부재에 작용할 때 발생하는 shear-plug 균열을 제어하기 위하여 충격하중이 가해지는 국부에 CFRP 전단보강을 실시하였다. 실험진행은 drop-weight test 방식으로 직접 기기를 만들어 실행하였다. 각각의 부재에 단계별로 충격하중을 가하여 실험을 진행하였으며 균열과 균열폭을 측정하였다. 실험결과 강섬유보강 RC보가 일반 RC보에 비하여 균열폭 및 shear-plug 균열제어 그리고 스폴링파괴에 더 높은 성능을 나타내었다. FRP로 부재의 하단을 휨 보강한 부재의 경우 균열의 제어에 어느정도 효과를 나타내고 있으나 충격하중이 가해질 시 콘크리트와 FRP 시트의 부착면에서 박리파괴가 빠르게 진행되었다. FRP 시트로 부재의 하단과 측면을 CFRP로 휨, 전단보강한 부재의 경우 shear-plug 균열제어에 가장 높은 저항성능을 보이고 있음을 확인할 수 있었다. 하지만 충격하중이 보강이 이루어지지 않은 부분에 작용할 경우 오히려 보강이 되지 않은 RC보에 비하여 콘크리트 스폴링 파괴에 더 취약함을 알 수 있었다.

• 한국터널지하공간학회 논문집
• /
• 제22권5호
• /
• pp.591-610
• /
• 2020

활성탄(Activated carbon) 및 이산화티탄(TiO₂)이 혼합된 콘크리트는 질소산화물(NOx) 저감에 있어 우수한 성능을 나타내기 때문에 지하공간 및 터널 내부의 미세먼지 저감 목적으로 활용되고 있다. 환경 및 구조물 노후 영향으로 터널 내부에 설치된 미세먼지 저감 콘크리트 표면에서 손상이 발생된다. 따라서 미세먼지 저감 콘크리트의 성능 유지를 위해 손상(박리) 유무평가가 필요하다. 본 연구에서는 전기비저항 특성을 이용하여 콘크리트 박리 유무 평가를 위한 기초연구를 수행하였다. 활성탄(0~15%) 및 TiO₂ (0~25%) 혼합비(시멘트 중량 기준)가 증가함에 따라 전기비저항 값은 감소하였다. 건조 조건에서 활성탄 및 TiO₂가 혼합된 시멘트 경화시편은 일반 시멘트 경화시편보다 전기비저항 값이 최대 2.3배 감소되었다. 또한, 포화 조건(포화도 = 85~98%)에서 활성탄만 혼합된 경화시편은 일반 시멘트 경화시편보다 전기비저항 값이 최대 3.5배 감소하는 결과를 보였다. 시편 상태(건조 또는 포화)와 관계없이 활성탄(15%) 및 TiO₂ (25%)가 혼합된 미세먼지 저감 시편의 경우, 일반 시멘트 시편과 비교하여 전기비저항 값은 약 2.3~2.8배 차이를 보였다. 본 연구결과는 전기비저항을 이용하여 터널 내 미세먼지 저감 콘크리트의 박리를 평가하기 위한 기초자료로 유용하게 활용될 것으로 기대된다.

• Corrosion Science and Technology
• /
• 제4권3호
• /
• pp.100-107
• /
• 2005

Reinforcing steel bars in reinforced concrete structures are protected from corrosion by passive film on the steel surface inside concrete with high alkalinity. However, when the passive film breaks down due to chloride ion ingressed into the RC structures, a corrosion initiates at the surface of steel bars. Then, internal pressure by volume expansion of corrosion products in reinforcing bars induces cracking and spalling of cover concrete, which reduces not only durability performance but also structural performance in RC structures. In this paper, a service life prediction of RC structures is carried out by using a micro-mechanics based corrosion model. The corrosion model is composed of a chloride penetration model to evaluate the initiation of corrosion and an electric corrosion cell model and an oxygen diffusion model to evaluate the rate and the accumulated amounts of corrosion. Then, a corrosion cracking model is combined to the models to evaluate critical amount of corrosion product for initiation cracking in cover concrete. By implementing the models into a finite element analysis program, a time and space dependent corrosion analysis and a service life prediction of RC structures due to chloride attack are simulated and the results of the analysis are compared with test results. The effect of crack width on the corrosion and the service life of the RC structures are analyzed and discussed.

• International Journal of Concrete Structures and Materials
• /
• 제1권1호
• /
• pp.75-81
• /
• 2007

The concrete pavement of the Seohae Highway in Korea has suffered from serious distress, only four to seven years after construction. Deterioration due to Alkali-Silica Reaction (ASR) has seldom been reported per se in Korea, because the aggregate used for the cement concrete has been considered safe against alkali-silica reaction so far. The purpose of this study is to examine the deterioration caused by an alkali-silica reaction of concrete pavement in Korea. The investigation methods included visual inspection and Automatic Road Analyzer (ARAN) analysis of surface cracks, coring for internal cracks, stereo microscopic analysis, scanning electronic microscope (SEM) analysis, and electron dispersive X-ray spectrometer (EDX) analysis. The results are presented as follows: the crack pattern of the concrete pavement in Korea was longitudinal cracking, map cracking or D-cracking. Local areas of damage were noticed four to five years after construction. The cracks started from edges or joints and spread out to slabs. The most intensive cracking was observed at the intersection of the transverse and longitudinal joints. Where cracking was the most intense, pieces of concrete and aggregate had spalled away from top surface and joint interface area. The progress of deterioration was very fast. The reaction product of alkali-silica gel was clearly identified by its generally colorless, white, or very pale yellow hue seen through a stereo optical microscopy. The typical locations of the reaction product were at the interface between aggregate and cement paste in a shape of a rim, within aggregate particles in the cracks, and in the large void in the cement paste. Most of the white products were found at interface or internal aggregates. SEM and EDX analysis confirmed that the white gel was a typical reaction product of ASR. The ASR gel in Korea mainly consisted of Silicate (Si) and Potassium (K) from the cement. The crack in the concrete pavement was caused by ASR. It seems that Korea is no longer safe from alkali-silica reaction.

• Advances in concrete construction
• /
• 제14권5호
• /
• pp.299-307
• /
• 2022

Cement-based matrixes have low tensile strength and negligible ductility. Adding fibres to these matrixes will improve their mechanical properties and make these composites suitable for structural applications. Post-cracking tensile strength of steel fibers-reinforced cementitious composite materials is directly related to the number of transverse fibers passing through the crack width and the pulling-out behavior of each of the fibers. Therefore, the exact recognition of the pullout behavior of single fibers is necessary to understand the uniaxial tensile and bending behavior of steel fiber-reinforced concrete. In this paper, an experimental study has been carried out on the pullout behavior of 3D (steel fibers with totally two hooks at both ends), 4D (steel fibers with a total of four hooks at both ends), and 5D (steel fibers with totally six hooks at both ends) in which the fibers have been located either perpendicular to the crack width or in an inclined manner. The pullout behavior of the mentioned steel fibers at an inclination angle of 0, 15, 30, 45, and 60 degrees and with embedded lengths of 10, 15, 20, 25, and 30 millimetres is studied in order to explore the simultaneous effect of the inclination angle of the fibers relative to the alongside loading and the embedded length of fibers on the pullout response in each case, including the maximal pullout force, the slip of the maximum point of pullout force, pullout energy, fiber rupture, and concrete matrix spalling. The results showed that the maximum pullout energy in 3D, 4D, and 5D steel fibers with different embedded lengths occurs at 0 to 30° inclination angles. In 5D fibers, maximum pullout energy occurs at a 30° angle with a 25 mm embedded length.

• Smart Structures and Systems
• /
• 제31권4호
• /
• pp.351-363
• /
• 2023

Visual structural inspections are an inseparable part of post-earthquake damage assessments. With unmanned aerial vehicles (UAVs) establishing a new frontier in visual inspections, there are major computational challenges in processing the collected massive amounts of high-resolution visual data. We propose twin deep learning models that can provide accurate high-resolution structural components and damage segmentation masks efficiently. The traditional approach to cope with high memory computational demands is to either uniformly downsample the raw images at the price of losing fine local details or cropping smaller parts of the images leading to a loss of global contextual information. Therefore, our twin models comprising Trainable Resizing for high-resolution Segmentation Network (TRS-Net) and DmgFormer approaches the global and local semantics from different perspectives. TRS-Net is a compound, high-resolution segmentation architecture equipped with learnable downsampler and upsampler modules to minimize information loss for optimal performance and efficiency. DmgFormer utilizes a transformer backbone and a convolutional decoder head with skip connections on a grid of crops aiming for high precision learning without downsizing. An augmented inference technique is used to boost performance further and reduce the possible loss of context due to grid cropping. Comprehensive experiments have been performed on the 3D physics-based graphics models (PBGMs) synthetic environments in the QuakeCity dataset. The proposed framework is evaluated using several metrics on three segmentation tasks: component type, component damage state, and global damage (crack, rebar, spalling). The models were developed as part of the 2^nd International Competition for Structural Health Monitoring.

• 한국안전학회지
• /
• 제32권6호
• /
• pp.98-103
• /
• 2017

In Korea, the guideline for the bridge safety inspection requests to assess surface degradation, including crack, efflorescence, spalling, and so on, for the rating of concrete bridges. Currently, the assessment of efflorescence is performed based on the visual inspection of expertized engineers, which may result in subjective inspection result. In this study, a novel method using a hyperspectral camera is proposed for objective and accurate assessment of concrete efflorescence. The hyperspectral camera acquires the light intensity for a number of continuous spectral bands of light for each pixel in an image, which makes the hyperspectral imaging technique provides more detailed information than a color camera that collects intensity for only three bands corresponding to RGB (red, green, and blue) colors. A stepwise assessment algorithm is proposed based on the spectral features to decompose efflorescence area from the inspected concrete area. The algorithm is tested in the laboratory test using two concrete specimens, one of which is dark colored with efflorescence on a surface while the other is bright concrete without efflorescence. The test shows high accuracy and applicability of the proposed efflorescence assessment using a hyperspectral camera.

• 한국분말재료학회지
• /
• 제21권1호
• /
• pp.44-49
• /
• 2014

In this study, nanocrystalline nickel powders were cold compacted by a dynamic compaction method using a single-stage gas gun system. A bending test was conducted to measure the bonding strengths of the compacted regions and microstructures of the specimen were analyzed using a scanning electron microscopy. The specimen was separated into two parts by a horizontal crack after compaction. Density test shows that the powder compaction occurred only in the upper part of the specimen. Brittle fracture was occurred during the bending test of the compact sample. Dispersion of shock energy due to spalling highly affected the bonding status of the nanocrystalline nickel powder.

• 한국터널지하공간학회 논문집
• /
• 제17권5호
• /
• pp.553-561
• /
• 2015

본 연구에서는 국내에 활발히 적용되기 시작한 쉴드 TBM 터널의 합리적인 유지관리를 위한 성능평가 기준을 제시하였다. 이를 위해 국내 외 성능평가 기준을 분석하였고, 국내 시공된 쉴드 TBM 터널에 대한 현장조사 및 정밀안전진단 보고서 분석을 통한 변상사례를 조사하여 성능평가 항목을 균열, 누수, 파손, 박리, 층분리 및 박락, 백태, 품질상태, 철근노출, 탄산화, 단차, 볼트상태, 배수상태, 지반상태, 접속부상태, 공동구 상태로 선정하였다. 또한 다중의사결정기법인 AHP 기법을 활용하여 선정된 성능평가에 대한 합리적인 가중치를 산정하였다.