• Nuclear Engineering and Technology
• /
• v.56 no.2
• /
• pp.558-567
• /
• 2024

Diagnostic support systems are being researched to assist operators in identifying and responding to abnormal events in a nuclear power plant. Most studies to date have considered single abnormal events only, for which it is relatively straightforward to obtain data to train the deep learning model of the diagnostic support system. However, cases in which multiple abnormal events occur must also be considered, for which obtaining training data becomes difficult due to the large number of combinations of possible abnormal events. This study proposes an approach to maintain diagnostic performance for multiple abnormal events by training a deep learning model with data on single abnormal events only. The proposed approach is applied to an existing algorithm that can perform feature selection and multi-label classification. We choose an extremely randomized trees classifier to select dedicated monitoring parameters for target abnormal events. In diagnosing each event occurrence independently, two-channel convolutional neural networks are employed as sub-models. The algorithm was tested in a case study with various scenarios, including single and multiple abnormal events. Results demonstrated that the proposed approach maintained diagnostic performance for 15 single abnormal events and significantly improved performance for 105 multiple abnormal events compared to the base model.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.27 no.3
• /
• pp.319-328
• /
• 2024

Semi-supervised learning is a good way to train a classification model using a small number of labeled and large number of unlabeled data. We applied semi-supervised learning to a synthetic aperture radar(SAR) image classification model with a limited number of datasets that are difficult to create. To address the previous difficulties, semi-supervised learning uses a model trained with a small amount of labeled data to generate and learn pseudo labels. Besides, a lot of number of papers use a single fixed threshold to create pseudo labels. In this paper, we present a semi-supervised synthetic aperture radar(SAR) image classification method that applies different thresholds for each class instead of all classes sharing a fixed threshold to improve SAR classification performance with a small number of labeled datasets.

• Structural Engineering and Mechanics
• /
• v.92 no.2
• /
• pp.121-131
• /
• 2024

This paper proposes a fault detection method for blade pitch systems of floating wind turbines using transformer-based deep-learning models. Transformers leverage self-attention mechanisms, efficiently process time-series data, and capture long-term dependencies more effectively than traditional recurrent neural networks (RNNs). The model was trained using normal operational data to detect anomalies through high reconstruction losses when encountering abnormal data. In this study, various fault conditions in a blade pitch system, including environmental load cases, were simulated using a detailed model of a spar-type floating wind turbine, the data collected from these simulations were used to train and test the transformer models. The model demonstrated superior fault-detection capabilities with high accuracy, precision, recall, and F1 scores. The results show that the proposed method successfully identifies faults and achieves high-performance metrics, outperforming existing traditional multi-layer perceptron (MLP) models and long short-term memory-autoencoder (LSTM-AE) models. This study highlights the potential of transformer models for real-time fault detection in wind turbines, contributing to more advanced condition-monitoring systems with minimal human intervention.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.28 no.10
• /
• pp.1603-1611
• /
• 2004

The validity of 'modified overlay model' to describe the cyclic behavior of annealed 316L stainless steel at room temperature was investigated. Material parameters(~f$_{i}$, m$_{i}$b, η, E) fur the model were obtained through constant strain amplitude test. The strain amplitude dependency of elastic limit and cyclic hardening, which were the characteristics of this model, were considered. Eight subelements were used to describe the nonlinearity of the hysteresis loops. The calculated hysteresis curve in each condition (0.5％, 0.7％, 0.9％ train amplitude test) was very close to the experimental one. Two tests, incremental step test and 5-step test, ere performed to check the validity of 'modified overlay model'. The elastic limit was saturated to the one of the highest strain amplitudes of the block in the incremental step test, so it seemed to be Masing material at the stabilized block. Cyclic hardening was successfully described in the increasing sequence of the strain amplitude in 5-step test. But, the slight cyclic softening followed by higher strain amplitude would not be able to simulate by'modified overlay model'. However, the discrepancy induced was very small between the calculated hystereses and the experimental ones. In conclusion,'Modified overlay model'was proved to be appropriate in strain range of 0.35％~ 1.0％..0％.

• Journal of Institute of Control, Robotics and Systems
• /
• v.7 no.8
• /
• pp.689-696
• /
• 2001

Large-scale Multi-disciplinary Systems(LMS) such as transportation, aerospace, defense etc. are complex systems in which there are many subsystems, interfaces, functions and demanding performance requirements. Because many contractors participate in the development, it is necessary to apply methods of sharing common objectives and communicating design status effectively among all of the stakeholders. The processes and methods of systems engineering which includes system requirement analysis; functional analysis; architecting; system analysis; interface control; and system specification development provide a success-oriented disciplined approach to the project. This paper shows not only the methodology and the results of model-based systems engineering to Automated Guided Transit(AGT) system as one of LMS systems, but also propose the extension of the model-based tool to help manage a project by linking WBS (Work Breakdown Structure), work organization, and PBS (Product Breakdown Structure). In performing the model-based functional analysis, the focus was on the operation concept of an example rail system at the top-level and the propulsion/braking function, a key function of the modern automated rail system. The model-based behavior analysis approach that applies a discrete-event simulation method facilitates the system functional definition and the test and verification activities. The first application of computer-aided tool, RDD-100, in the railway industry demonstrates the capability to model product design knowledge and decisions concerning key issues such as the rationale for architecting the top-level system. The model-based product design knowledge will be essential in integrating the follow-on life-cycle phase activities. production through operation and support, over the life of the AGT system. Additionally, when a new generation train system is required, the reuse of the model-based database can increase the system design productivity and effectiveness significantly.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39A no.8
• /
• pp.437-444
• /
• 2014

In this paper, we propose a path loss model with the multiple antennas and doppler shift for high speed railroad communication. Path loss model is very important in order to design consider diverse characteristic in high-speed train communication. Currently wireless communication systems use the multiple antennas in order to improve the channel capacity or diversity gain. However, until recently, many researches on path loss model only consider geographical environment between the transmitter and the receiver. There is no study about path loss model considering diversity effect and doppler shift. In order to make average residuals considering doppler shift we use tuned free space path loss model which is utilized for measurement results at high speed railroad. The environment of high speed rail is mostly at viaduct and flatland over than 50 percent. And in order to make average residuals considering multiple antenna we use theoretical estimation of diversity gain with MRC scheme. proposed model predict loss of received signal by estimating average residuals between diversity effect and doppler shift.

• Structural Engineering and Mechanics
• /
• v.57 no.4
• /
• pp.657-680
• /
• 2016

Reinforced concrete (RC) deep beams are structural members that predominantly fail in shear. Therefore, determining the shear strength of these types of beams is very important. The strut-and-tie method is commonly used to design deep beams, and this method has been adopted in many building codes (ACI318-14, Eurocode 2-2004, CSA A23.3-2004). In this study, the efficiency of artificial neural networks (ANNs) in predicting the shear strength of RC deep beams is investigated as a different approach to the strut-and-tie method. An ANN model was developed using experimental data for 214 normal and high-strength concrete deep beams from an existing literature database. Seven different input parameters affecting the shear strength of the RC deep beams were selected to create the ANN structure. Each parameter was arranged as an input vector and a corresponding output vector that includes the shear strength of the RC deep beam. The ANN model was trained and tested using a multi-layered back-propagation method. The most convenient ANN algorithm was determined as trainGDX. Additionally, the results in the existing literature and the accuracy of the strut-and-tie model in ACI318-14 in predicting the shear strength of the RC deep beams were investigated using the same test data. The study shows that the ANN model provides acceptable predictions of the ultimate shear strength of RC deep beams (maximum $R^2{\approx}0.97$). Additionally, the ANN model is shown to provide more accurate predictions of the shear capacity than all the other computed methods in this study. The ACI318-14-STM method was very conservative, as expected. Moreover, the study shows that the proposed ANN model predicts the shear strengths of RC deep beams better than does the strut-and-tie model approaches.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.14 no.4
• /
• pp.18-26
• /
• 2015

This study is a fundamental research to suggest a forecasting model for short-term railway passenger demand focusing on major lines (Gyeungbu, Honam, Jeonla, Janghang, Jungang) of Saemaeul rail and Mugunghwa rail. Also the author tried to verify the potential application of the proposed models. For this study, SARIMA model considering characteristics of seasonal trip is basically used, and daily mean forecasting models are independently constructed depending on weekday/weekend in order to consider characteristics of weekday/weekend trip and a legal holiday trip. Furthermore, intervention events having an impact on using the train such as introduction of new lines or EXPO are reflected in the model to increase reliability of the model. Finally, proposed models are confirmed to have high accuracy and reliability by verifying predictability of models. The proposed models of this research will be expected to utilize for establishing a plan for short-term operation of lines.

• Korean Journal of Remote Sensing
• /
• v.38 no.1
• /
• pp.111-126
• /
• 2022

Land use and land cover (LULC) mapping is an important factor in geospatial analysis. Although highly precise ground-based LULC monitoring is possible, it is time consuming and costly. Conversely, because the synthetic aperture radar (SAR) sensor is an all-weather sensor with high resolution, it could replace field-based LULC monitoring systems with low cost and less time requirement. Thus, LULC is one of the major areas in SAR applications. We developed a LULC model using only KOMPSAT-5 single co-polarized data and digital elevation model (DEM) data. Twelve HH-polarized images and 18 VV-polarized images were collected, and two HH-polarized images and four VV-polarized images were selected for the model testing. To train the LULC model, we applied the conditional generative adversarial network (cGAN) method. We used U-Net combined with the residual unit (ResUNet) model to generate the cGAN method. When analyzing the training history at 1732 epochs, the ResUNet model showed a maximum overall accuracy (OA) of 93.89 and a Kappa coefficient of 0.91. The model exhibited high performance in the test datasets with an OA greater than 90. The model accurately distinguished water body areas and showed lower accuracy in wetlands than in the other LULC types. The effect of the DEM on the accuracy of LULC was analyzed. When assessing the accuracy with respect to the incidence angle, owing to the radar shadow caused by the side-looking system of the SAR sensor, the OA tended to decrease as the incidence angle increased. This study is the first to use only KOMPSAT-5 single co-polarized data and deep learning methods to demonstrate the possibility of high-performance LULC monitoring. This study contributes to Earth surface monitoring and the development of deep learning approaches using the KOMPSAT-5 data.

• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.2
• /
• pp.46.3-47
• /
• 2021

Recently a multilayer spectral inversion (MLSI) model has been proposed to infer the physical parameters of plasmas in the solar chromosphere. The inversion solves a three-layer radiative transfer model using the strong absorption line profiles, H alpha and Ca II 8542 Å, taken by the Fast Imaging Solar Spectrograph (FISS). The model successfully provides the physical plasma parameters, such as source functions, Doppler velocities, and Doppler widths in the layers of the photosphere to the chromosphere. However, it is quite expensive to apply the MLSI to a huge number of line profiles. For example, the calculating time is an hour to several hours depending on the size of the scan raster. We apply deep neural network (DNN) to the inversion code to reduce the cost of calculating the physical parameters. We train the models using pairs of absorption line profiles from FISS and their 13 physical parameters (source functions, Doppler velocities, Doppler widths in the chromosphere, and the pre-determined parameters for the photosphere) calculated from the spectral inversion code for 49 scan rasters (~2,000,000 dataset) including quiet and active regions. We use fully connected dense layers for training the model. In addition, we utilize a skip connection to avoid a problem of vanishing gradients. We evaluate the model by comparing the pairs of absorption line profiles and their inverted physical parameters from other quiet and active regions. Our result shows that the deep learning model successfully reproduces physical parameter maps of a scan raster observation per second within 15% of mean absolute percentage error and the mean squared error of 0.3 to 0.003 depending on the parameters. Taking this advantage of high performance of the deep learning model, we plan to provide the physical parameter maps from the FISS observations to understand the chromospheric plasma conditions in various solar features.