• Structural Engineering and Mechanics
• /
• 제77권1호
• /
• pp.47-56
• /
• 2021

This paper deals with damage detection using a Gapped Smoothing Method (GSM) combined with deep learning. Convolutional Neural Network (CNN) is a model of deep learning. CNN has an input layer, an output layer, and a number of hidden layers that consist of convolutional layers. The input layer is a tensor with shape (number of images) × (image width) × (image height) × (image depth). An activation function is applied each time to this tensor passing through a hidden layer and the last layer is the fully connected layer. After the fully connected layer, the output layer, which is the final layer, is predicted by CNN. In this paper, a complete machine learning system is introduced. The training data was taken from a Finite Element (FE) model. The input images are the contour plots of curvature gapped smooth damage index. A free-free beam is used as a case study. In the first step, the FE model of the beam was used to generate data. The collected data were then divided into two parts, i.e. 70% for training and 30% for validation. In the second step, the proposed CNN was trained using training data and then validated using available data. Furthermore, a vibration experiment on steel damaged beam in free-free support condition was carried out in the laboratory to test the method. A total number of 15 accelerometers were set up to measure the mode shapes and calculate the curvature gapped smooth of the damaged beam. Two scenarios were introduced with different severities of the damage. The results showed that the trained CNN was successful in detecting the location as well as the severity of the damage in the experimental damaged beam.

• 한국복합재료학회:학술대회논문집
• /
• 한국복합재료학회 2003년도 춘계학술발표대회 논문집
• /
• pp.125-128
• /
• 2003

Low-velocity impact damage is a major concern in the design of structures made of composite materials, because impact damage is hidden inside and cannot be detected by visual inspection. The piezoelectric thin film sensor can be used to detect variations in structural and material properties for structural health monitoring. In this paper, the PVDF and PZT sensors were used for monitoring impact damage initiation in Gr/Ep composite panel to illustrate this potential benefit. A series of impact test at various impact energy by changing impact mass and height is performed on the instrumented drop weight impact tester. The wavelet transform(WT) is used to decompose the piezoelectric sensor signals in this study. Test results show that the particular waveform of sensor signals implying the damage initiation and development are detected above the damage initiation impact energy. And it is found that both PZT and PVDF sensors can be used to detect the impact damage.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2018년도 학술발표회
• /
• pp.499-499
• /
• 2018

Hydro-meteorological extremes are trivial in these days. Therefore, it is important to identify extreme hydrological events in advance to mitigate the damage due to the extreme events. In this context, exploring temporal distribution of sub-daily extreme rainfall at multiple rain gauges would informative to identify different states to describe severity of the disaster. This study proposehidden Markov chain model (HMM) based rainfall analysis tool to understand the temporal sub-daily rainfall patterns over South Korea. Hourly and daily rainfall data between 1961 and 2017 for 92 stations were used for the study. HMM was applied to daily rainfall series to identify an observed hidden state associated with rainfall frequency and intensity, and further utilized the estimated hidden states to derive a temporal distribution of daily extreme rainfall. Transition between states over time was clearly identified, because HMM obviously identifies the temporal dependence in the daily rainfall states. The proposed HMM was very useful tool to derive the temporal attributes of the daily rainfall in South Korea. Further, daily rainfall series were disaggregated into sub-daily rainfall sequences based on the temporal distribution of hourly rainfall data.

• 한국전산구조공학회논문집
• /
• 제16권3호
• /
• pp.229-236
• /
• 2003

손상평가를 위해 많은 연구자들에 의해 인공신경망이 이용되어 왔다. 그러나, 인공신경망을 이용한 손상평가에 있어 정확성과 능률성을 제고하기 위해서는 몇가지 문제점이 있다. 기존의 인공신경망 특히 역전파신경망(BPNN)의 경우 신경망 학습을 위해 많은 수의 학습패턴을 필요로 하며, 또한 신경망의 구조와 해의 수렴간에 어떤 확정적인 관계가 존재하지 않는다. 따라서 신경망의 은닉층의 수와 한 은닉층에서의 노드수는 시행착오적으로 결정되게 된다. 이러한 많은 훈련패턴의 준비와 최적의 신경망 구조 결정을 위해서는 많은 시간이 필요하다. 본 논문에서는 이러한 단점들을 극복하기 위해 확률신경망을 패턴분류기로 사용하였다. 이를 판형철도교의 손상평가에 수치해석적으로 검증하였다. 또한 확률신경망을 이용한 철도판형교 손상평가시 적절한 훈련패턴 선택을 위해 모드형상과 고유진동수를 사용한 경우의 적용성에 대해 검토하였다.

• Structural Monitoring and Maintenance
• /
• 제5권4호
• /
• pp.507-519
• /
• 2018

In this paper, damage assessment in wind-turbine towers using vibration-based artificial neural networks (ANNs) is numerically investigated. At first, a vibration-based ANNs algorithm is designed for damage detection in a wind turbine tower. The ANNs architecture consists of an input, an output, and hidden layers. Modal parameters of the wind turbine tower such as mode shapes and frequencies are utilized as the input and the output layer composes of element stiffness indices. Next, the finite element model of a real wind-turbine tower is established as the test structure. The natural frequencies and mode shapes of the test structure are computed under various damage cases of single and multiple damages to generate training patterns. Finally, the ANNs are trained using the generated training patterns and employed to detect damaged elements and severities in the test structure.

• Environmental Analysis Health and Toxicology
• /
• 제27권
• /
• pp.1.1-1.3
• /
• 2012

• 국제학술발표논문집
• /
• The 8th International Conference on Construction Engineering and Project Management
• /
• pp.505-509
• /
• 2020

Twitter is a useful medium to grasp various damage situations that have occurred in society. However, it is a laborious task to spot damage-related topics according to time in the environment where information is constantly produced. This paper proposes a methodology of constructing a knowledge structure by combining the BERT-based classifier and the community detection techniques to discover the topics underlain in the damage information. The methodology consists of two steps. In the first step, the tweets are classified into the classes that are related to human damage, infrastructure damage, and industrial activity damage by a BERT-based transfer learning approach. In the second step, networks of the words that appear in the damage-related tweets are constructed based on the co-occurrence matrix. The derived networks are partitioned by maximizing the modularity to reveal the hidden topics. Five keywords with high values of degree centrality are selected to interpret the topics. The proposed methodology is validated with the Hurricane Harvey test data.

• Steel and Composite Structures
• /
• 제46권3호
• /
• pp.319-334
• /
• 2023

Localizing damages is an essential task to monitor the health of the structures since they may not be able to operate anymore. Among the damage detection techniques, non-destructive methods are considerably more preferred than destructive methods since damage can be located without affecting the structural integrity. However, these methods have several drawbacks in terms of detecting abilities, time consumption, cost, and hardware or software requirements. Employing artificial intelligence techniques could overcome such issues and could provide a powerful damage detection model if the technique is utilized correctly. In this study, the crack localization in flat and folded plate structures has been conducted by employing a Backpropagated Artificial Neural Network (BPANN). For this purpose, cracks with 18 different dimensions in thin, flat, and folded structures having 15⁰, 30⁰, 45⁰, and 60⁰ folding angle have been modeled and subjected to free vibration analysis by employing the Classical Plate Theory with Finite Element Method. A Four-nodded quadrilateral element having six degrees of freedom has been considered to represent those structures mathematically. The first ten natural frequencies have been obtained regarding healthy and cracked structures. To localize the crack, the ratios of the frequencies of the cracked flat and folded structures to those of healthy ones have been taken into account. Those ratios have been given to BPANN as the input variables, while the crack locations have been considered as the output variables. A total of 500 crack locations have been regarded within the dataset obtained from the results of the free vibration analysis. To build the best intelligent model, a feature search has been conducted for BAPNN regarding activation function, the number of hidden layers, and the number of hidden neurons. Regarding the analysis results, it is concluded that the BPANN is able to localize the cracks with an average accuracy of 95.12%.

• Smart Structures and Systems
• /
• 제15권4호
• /
• pp.997-1018
• /
• 2015

Structural health monitoring along with damage detection and assessment of its severity level in non-accessible reinforced concrete members using piezoelectric materials becomes essential since engineers often face the problem of detecting hidden damage. In this study, the potential of the detection of flexural damage state in the lower part of the mid-span area of a simply supported reinforced concrete beam using piezoelectric sensors is analytically investigated. Two common severity levels of flexural damage are examined: (i) cracking of concrete that extends from the external lower fiber of concrete up to the steel reinforcement and (ii) yielding of reinforcing bars that occurs for higher levels of bending moment and after the flexural cracking. The purpose of this investigation is to apply finite element modeling using admittance based signature data to analyze its accuracy and to check the potential use of this technique to monitor structural damage in real-time. It has been indicated that damage detection capability greatly depends on the frequency selection rather than on the level of the harmonic excitation loading. This way, the excitation loading sequence can have a level low enough that the technique may be considered as applicable and effective for real structures. Further, it is concluded that the closest applied piezoelectric sensor to the flexural damage demonstrates higher overall sensitivity to structural damage in the entire frequency band for both damage states with respect to the other used sensors. However, the observed sensitivity of the other sensors becomes comparatively high in the peak values of the root mean square deviation index.

• 한국전산구조공학회논문집
• /
• 제33권6호
• /
• pp.359-366
• /
• 2020

본 논문에서는 전산구조 해석 데이터를 기반으로 인공 신경망을 활용하여 헬리데크 구조물에 대한 손상 추정 기법을 제안하였다. 헬리데크를 구성하는 트러스와 서포트 부재들에 대해서 절점을 공유하는 부재들을 70개의 모델로 그룹화 하였으며, 최대 3가지 부재 그룹에 무작위로 손상을 부여하여 총 37,400개의 손상 시나리오를 생성하였다. 이들 각각에 대해서 구조 해석 프로그램을 통해 모드 해석을 수행하였으며, 전체 손상 시나리오를 사용 목적에 따라 학습, 유효성 검사, 그리고 검증 시나리오로 분리하였다. 헬리데크의 손상 및 비손상 상태의 동적 응답 특성에 대한 패턴 인식을 위해 PyTorch 프로그램을 활용하여 3개의 은닉층을 가지는 인공 신경망을 구성하였으며, 이에 대해서 다양한 손상 시나리오를 반복 학습함으로써 손실 함수를 최소로 하는 인공 신경망을 도출하였다. 최종적으로 총 400개의 검증 시나리오에 대해서 인공 신경망이 추정한 손상률과 실제 부여된 손상률을 비교하였으며, 그 결과 본 연구를 통해 얻어진 인공 신경망이 손상 부재의 위치와 손상 정도를 매우 높은 정확도로 예측하는 것을 확인하였다.