https://doi.org/10.12652/Ksce.2025.45.4.0431

김동우(Kim, Dongwoo);서병훈(Seo, Byunghun);김동수(Kim, Dongsu);조예림(Jo, Yerim);이종혁(Lee, Jonghyuk);최원(Choi, Won)

With the increasing frequency and severity of extreme weather events such as strong winds and heavy snowfall due to climate change, structural damage to plastic greenhouses in Korea has also been on the rise. To mitigate such damage in advance, this study developed a Transformer-based time series forecasting model to predict snow depth and wind speed, and integrated the predictions with a nonlinear structural analysis model to estimate greenhouse safety factors. The proposed model was designed to forecast snow depth and wind speed 1 to 12 hours ahead based on the previous 6 hours of meteorological data, and its hyperparameters were optimized using Bayesian optimization. The model demonstrated high accuracy in predicting snow depth, with RMSE ranging from 0.66 to 3.78 cm, while the prediction performance for wind speed showed R2 values between 0.17 and 0.67. Comparative analysis with an LSTM model under the same conditions revealed that the Transformer model was less sensitive to the non-stationary nature of time series data and exhibited more stable performance in long-term forecasting. Finally, the model was applied to actual damage cases caused by heavy snowfall, and the resulting safety factors from structural analysis were found to be correlated with the observed damage area. This study provides a technical foundation for the preemptive prediction and response to greenhouse damage caused by heavy snowfall and is expected to contribute meaningfully to reducing agricultural losses in the future.

https://doi.org/10.12652/Ksce.2025.45.4.0445

원치문(Won, Chi Moon)

This paper presents the effect of neglecting the self-weight on the natural frequencies of simply supported [αββααβ]r type composite laminated plates with attached masses. The presented method utilizes deflection influence surface which can be obtained by any calculation methods, and determines the deflection mode shapes of the structure under resonance conditions based on inertial forces. A parametric study is conducted to examine the influence of several factors, including the number of laminates, fiber orientation, and both the magnitude and location of the attached masses, on the natural frequencies of the laminated plates when self-weight is neglected. The analysis results show that the ratio of natural frequencies with and without considering self-weight is scarcely affected by the number of laminates, but is significantly influenced by the magnitude and location of the attached mass. Notably, the influence of fiber orientation varied depending on the location of the attached mass, and when the attached mass was positioned equidistance from the x and y axes, the fiber orientation had no effect on the natural frequency ratio. Also, the closer the attached mass is to the center of the plate, the more rapidly the natural frequency without considering self-weight converges to that with self-weight. Based on these results, it is possible to estimate the actual natural frequencies from those calculated by a simple method neglecting self-weight in case of where attached masses are present. The results of this study can serve as a fundamental reference for the dynamic analysis and design of composite laminated plates with attached masses.

https://doi.org/10.12652/Ksce.2025.45.4.0459

김주훈(Kim. Joo-hun);최윤석(Choi, Yun-seok)

The purpose of this study was to estimate the flood volume and inundation extent in the Sinuiju-Si that occured in July 2024 using satellite-derived precipitation data and global terrain data. The precipitation data for flood volume analysis used JAXA's GSMaP data, and the terrain data for inundation extent analysis used JAXA's AW3D30 DSM data. The tools for flood discharge analysis and inundation analysis used GRM and G2D models, respectively. The satellite-derived precipitation data can be effectively used in areas such as North Korea where it is difficult to collect measured data and the data's spatiotemporal resolution is low. As a result of the flood volume simulation, the peak flow downstream of Taepyeong Dam was 92,574.5 ㎥/s, and the occurrence time was analyzed as 22:00 on July 28. As a result of the simulation using the G2D model, the maximum inundation extent was analyzed as 05:00 on July 29. It is estimated that approximately 67 % of the area of Sinuiju-Si was flooded. In particular, it can be confirmed that most of the flood was concentrated in the Sinuiju-Si area of North Korea, which was due to the topographical differences between Sinuiju-Si in North Korea and Dandong in China. In flood simulation of this study, the accuracy of DEM/DSM has a great influence on the accuracy of the simulation results, but through flood simulation, similar results were obtained in terms of flood depth and flood range as reported in the media at the time of the flood, and the changes in flood depth and flood time in the Sinuiju-Si area could be quantitatively analyzed. The method applied in this study can be usefully utilized to quantitatively evaluate the flood scale in unmeasured areas in the future.

https://doi.org/10.12652/Ksce.2025.45.4.0469

이동민(Lee, Dongmin);오랑치맥 솜야(Uranchimeg, Sumiya);권현한(Kwon, Hyun-Han)

This study develops a machine learning-based inflow forecasting model by reconstructing input time series using a state-space embedding method that captures nonlinear and nonstationary behaviors. The target sites are the Hapcheon Dam and Namgang Dam in the Nakdong River Basin, with areal rainfall, upstream water level, and inflow data used as model inputs. The delay time and embedding dimension were determined using the Average Mutual Information (AMI) and False Nearest Neighbor (FNN) methods. Long Short-Term Memory (LSTM) and Support Vector Machine (SVM) models were applied to predict inflows with lead times ranging from 1 to 6 hours. For the Hapcheon Dam, the SVM model achieved a 6-hour lead time prediction with a correlation coefficient(CC) of 0.887 and an RMSE of 88.9 m3/s, significantly outperforming LSTM (CC = 0.655, RMSE = 234.51 m3/s). For the Namgang Dam, the 6-hour prediction showed a smaller performance difference between SVM (CC = 0.887) and LSTM (CC = 0.858), but SVM tended to underestimate peak inflows. These results show that SVM performs well under normal to moderate inflow conditions, while LSTM is more effective at capturing extreme peaks. This study presents a practical inflow forecasting framework by combining optimized input reconstruction with machine learning, and future improvements are anticipated through integrating weather forecasts and spatial data.

https://doi.org/10.12652/Ksce.2025.45.4.0483

최준우(Choi, Junwoo)

To address the growing risk of coastal flooding driven by sea level rise and increasingly frequent extreme storm surges, a preliminary study was conducted using the phase-resolving Boussinesq model FUNWAVE-TVD. A newly developed module incorporating river discharge and precipitation was implemented to simulate hydrodynamic processes in the coastal and estuarine region surrounding the Mispillion River in Delaware Bay, eastern United States. The study examined the influence of compound flooding on estuarine water levels. Results showed that elevated coastal water levels increased water stages in the estuary and lower river reach, inducing flow retardation and reversal. Simultaneous direct inundation from storm surge and inland inundation through the river channel were observed. In downstream and low-lying areas, the superposition of multiple forcings intensified inundation extent. Additionally, drainage limitations caused prolonged flooding even after coastal water levels subsided. This test highlights the applicability of the enhanced FUNWAVE-TVD framework for simulating compound flooding in coastal and estuarine systems and emphasizes the importance of riverine inputs in phase-resolving models.

https://doi.org/10.12652/Ksce.2025.45.4.0493

도명식(Do, Myungsik);정유미(Jeong, Yumi)

This study aims to evaluate the quality and productivity of smart construction through empirical analysis utilizing construction automation equipment and to propose strategies for promoting the on-site application of such equipment. The analysis was conducted on major earthwork operations, including slope cutting, pipeline trenching, and soil transport, by comparing the performance of conventional excavators with MG·MC (Machine Guidance·Machine Control) manned and remote-controlled excavators, and traditional dump trucks with autonomous haulers. The results showed that MG·MC manned and remote-controlled excavators demonstrated superior quality and productivity compared to conventional excavators, effectively reducing labor dependency on construction sites. In particular, MG·MC manned excavators achieved higher work accuracy and productivity than standard unit cost estimates, confirming that digital and automation technologies can significantly contribute to improving safety, enhancing operational efficiency, and reducing labor costs. Although autonomous haulers have limited payload capacity, they achieved a sufficient level of productivity and demonstrated strong potential to replace conventional dump trucks as technology advances. This empirical study confirms the feasibility and effectiveness of automation technologies in real-world applications. To further promote the adoption of smart construction technologies, policy support and continuous technological development are required, along with efforts to address technical limitations and legal constraints.

https://doi.org/10.12652/Ksce.2025.45.4.0505

김창윤(Kim, Changyoon);정경원(Jung, Kyungwon);이창준(Lee, Changjun)

In alignment with the global movement toward carbon neutrality, South Korea has actively pursued the transition to renewable energy sources, such as solar and wind power, with strong government support. However, the implementation of these projects has sometimes been hasty, leading to conflicts and opposition from local residents in areas where renewable energy facilities were installed. To address this issue, the government introduced policies like community-based renewable energy projects, which aim to enhance local acceptance by expanding benefit-sharing with residents. Despite these efforts, challenges in project execution have emerged due to insufficient communication and a lack of transparency among stakeholders, resulting in delays or cancellations of some projects. This study proposes the application of Shapley values, commonly used in cooperative game theory for resource allocation, to analyze the contributions of different participants and the distribution of benefits in community-based renewable energy projects. By calculating the Shapley value, the study quantifies the contribution of each participant to the project's success, enabling a fair distribution of profits based on relative contributions. This approach, by ensuring a transparent and equitable benefit-sharing process, is expected to reduce conflicts among stakeholders and strengthen acceptance by local communities and municipalities.

https://doi.org/10.12652/Ksce.2025.45.4.0513

박영훈(Park. Younghoon)

This study aims to effectively analyze damage images of reinforced concrete structures by comparing the performance of various convolutional neural network backbone models based on transfer learning, in order to identify the optimal architecture. A total of 3,500 damage images were used in experiments involving 12 pretrained models and 16 combinations of hyperparameters. The results showedthat while the highest top-1 validation accuracy of a vanilla CNN was limited to 67.5 %, the accuracy significantly improved to 86.0 % when using the EfficientNetB7 model with transfer learning, clearly demonstrating the benefit of transfer learning in this domain. However, classification performance cannot be fully evaluated by accuracy alone. In terms of F1-score, the InceptionV3 model exhibited the most balanced performance. Although the MobileNet series showed excellent efficiency and lightweight characteristics, it had limitations in precise classification under class-imbalanced conditions. Since high-precision classification is essential for structural damage image analysis, this study confirms the importance of considering comprehensive performance metrics such as F1-score in addition to accuracy. Furthermore, the findings suggest that additional validation under diverse environmental conditions and improvements in real-time processing capabilities are necessary for practical applications. Future work should focus on addressing data imbalance, enhancing the performance of lightweight models, and developing optimization techniques.