• Structural Engineering and Mechanics
• /
• v.52 no.5
• /
• pp.1019-1032
• /
• 2014

Rails are key elements in railway superstructure since these elements receive directly the train load transmitted by the wheels. Simultaneously, rails must provide effective stress transference to the rest of the track elements. This track element often deteriorates as a consequence of the vehicle passing or manufacturing imperfections that cause in rail several defects. Among these rail defects, transverse cracks highlights and are considered a severe pathology because they can suddenly trigger the rail failure. This study is focused on UIC-60 rails with transverse cracks. A 3-D FEM model is developed in ANSYS for the flawless rail in which conditions simulating the crack presence are implemented. To account for the inertia loss of the rail as a consequence of the cracking, a reduction of the bending stiffness of the rail is considered. The numerical models have been calibrated using the first four bending vibration modes in terms of frequencies. These vibration frequencies have been obtained using the Experimental Modal Analysis technique, studying the changes in the modal parameters of the rails induced by the crack and comparing the results obtained by the model with experimental results. Finally, the calibrated and validated models for the single rail have been implemented in a complete railway ballasted track FEM model in order to study the static influence of the cracks on the rail deflection caused by a load passing.

• The Transactions of the Korean Institute of Electrical Engineers A
• /
• v.50 no.5
• /
• pp.248-254
• /
• 2001

The traction power demand highly varies with time and train positions and the traction load is a large-capacity current at single phase converted from 3-phase power system. Subsequently, each phase current converted from 3-phase power system cannot be maintained in balance any longer and thus the traction load can bring about imbalance in three-phase voltage. Therefore, the exact assessment of voltage unbalance must be carried out preferentially as well as load forecast at stages of designing and planning for electric railway system. The evaluation of unbalance voltage in areas, such as electric railway depots should be a prerequisite with more accuracy. The conventional researches on voltage unbalance have dealt with connection schemes of the transformers used in ac AT-fed electric railroads system and induced formulas to briefly evaluate voltage unbalance in the system(3). These formulas are still being used widely due to their easy applicabilities on voltage unbalance evaluation. Meanwhile, they don't take into account detailed characteristics of ac AT-fed electric railroads system, being founded on some assumptions. Accordingly. accuracy still remains in question. This paper proposes a new method to more effectively estimate voltage unbalance index. In this method, numerous diverted circuits in electric railway depots are categorized in three components and each component is defined as a two-port network model. The equivalent circuit for the entire power supply system is also described into a two-port network model by making parallel and/or series connections of these components. Efficiency and accuracy in voltage unbalance calculation as well can be promoted by simplifying the circuits into two-port network models.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.16 no.3
• /
• pp.401-406
• /
• 2021

Recently, mean-teacher models based on convolutional recurrent neural networks are popularly used in audio event detection. The mean-teacher model is an architecture that consists of two parallel CRNNs and it is possible to train them effectively on the weakly-labelled and unlabeled audio data by using the consistency learning metric at the output of the two neural networks. In this study, we tried to improve the performance of the mean-teacher model by using additional derivative features of the log-mel spectrum. In the audio event detection experiments using the training and test data from the Task 4 of the DCASE 2018/2019 Challenges, we could obtain maximally a 8.1% relative decrease in the ER(Error Rate) in the mean-teacher model using proposed derivative features.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.12
• /
• pp.45-52
• /
• 2021

In this paper, we propose a prediction model for osteopenia and osteoporosis based on a convolutional neural network(CNN) using computed tomography(CT) images. In a single CT image, CNN had a limitation in utilizing important local features for diagnosis. So we propose a compound model which has two identical structures. As an input, two different texture images are used, which are converted from a single normalized CT image. The two networks train different information by using dissimilarity loss function. As a result, our model trains various features in a single CT image which includes important local features, then we ensemble them to improve the accuracy of predicting osteopenia and osteoporosis. In experiment results, our method shows an accuracy of 77.11% and the feature visualize of this model is confirmed by using Grad-CAM.

• Genomics & Informatics
• /
• v.19 no.1
• /
• pp.11.1-11.8
• /
• 2021

For the novel coronavirus disease 2019 (COVID-19), predictive modeling, in the literature, uses broadly susceptible exposed infected recoverd (SEIR)/SIR, agent-based, curve-fitting models. Governments and legislative bodies rely on insights from prediction models to suggest new policies and to assess the effectiveness of enforced policies. Therefore, access to accurate outbreak prediction models is essential to obtain insights into the likely spread and consequences of infectious diseases. The objective of this study is to predict the future COVID-19 situation of Korea. Here, we employed 5 models for this analysis; SEIR, local linear regression (LLR), negative binomial (NB) regression, segment Poisson, deep-learning based long short-term memory models (LSTM) and tree based gradient boosting machine (GBM). After prediction, model performance comparison was evelauated using relative mean squared errors (RMSE) for two sets of train (January 20, 2020-December 31, 2020 and January 20, 2020-January 31, 2021) and testing data (January 1, 2021-February 28, 2021 and February 1, 2021-February 28, 2021) . Except for segmented Poisson model, the other models predicted a decline in the daily confirmed cases in the country for the coming future. RMSE values' comparison showed that LLR, GBM, SEIR, NB, and LSTM respectively, performed well in the forecasting of the pandemic situation of the country. A good understanding of the epidemic dynamics would greatly enhance the control and prevention of COVID-19 and other infectious diseases. Therefore, with increasing daily confirmed cases since this year, these results could help in the pandemic response by informing decisions about planning, resource allocation, and decision concerning social distancing policies.

• The Bulletin of The Korean Astronomical Society
• /
• v.44 no.1
• /
• pp.51.3-51.3
• /
• 2019

Solar magnetograms are important for studying solar activity and predicting space weather disturbances1. Farside magnetograms can be constructed from local helioseismology without any farside data2-4, but their quality is lower than that of typical frontside magnetograms. Here we generate farside solar magnetograms from STEREO/Extreme UltraViolet Imager (EUVI) $304-{\AA}$ images using a deep learning model based on conditional generative adversarial networks (cGANs). We train the model using pairs of Solar Dynamics Observatory (SDO)/Atmospheric Imaging Assembly (AIA) $304-{\AA}$ images and SDO/Helioseismic and Magnetic Imager (HMI) magnetograms taken from 2011 to 2017 except for September and October each year. We evaluate the model by comparing pairs of SDO/HMI magnetograms and cGAN-generated magnetograms in September and October. Our method successfully generates frontside solar magnetograms from SDO/AIA $304-{\AA}$ images and these are similar to those of the SDO/HMI, with Hale-patterned active regions being well replicated. Thus we can monitor the temporal evolution of magnetic fields from the farside to the frontside of the Sun using SDO/HMI and farside magnetograms generated by our model when farside extreme-ultraviolet data are available. This study presents an application of image-to-image translation based on cGANs to scientific data.

• Journal of Convergence for Information Technology
• /
• v.12 no.1
• /
• pp.31-38
• /
• 2022

This paper applies an expert independent unsupervised neural network learning-based multivariate time series data analysis model, MSCRED(Multi-Scale Convolutional Recurrent Encoder-Decoder), and to overcome the limitation, because the MCRED is based on Auto-encoder model, that train data must not to be contaminated, by using learning data sampling technique, called Subset Sampling Validation. By using the vibration data of power plant equipment that has been labeled, the classification performance of MSCRED is evaluated with the Anomaly Score in many cases, 1) the abnormal data is mixed with the training data 2) when the abnormal data is removed from the training data in case 1. Through this, this paper presents an expert-independent anomaly diagnosis framework that is strong against error data, and presents a concise and accurate solution in various fields of multivariate time series data.

• Journal of Korea Water Resources Association
• /
• v.55 no.7
• /
• pp.495-504
• /
• 2022

With the recent development of artificial intelligence, a Long Short-Term Memory (LSTM) model that is efficient with time-series analysis is being used to increase the accuracy of predicting the inflow of dams. In this study, we predict the inflow of the Soyang River dam, using the LSTM model with the Sequence-to-Sequence (LSTM-s2s) and attention mechanism (LSTM-s2s with attention) that can further improve the LSTM performance. Hourly inflow, temperature, and precipitation data from 2013 to 2020 were used to train the model, and validate and test for evaluating the performance of the models. As a result, the LSTM-s2s with attention showed better performance than the LSTM-s2s in general as well as in predicting a peak value. Both models captured the inflow pattern during the peaks but detailed hourly variability is limitedly simulated. We conclude that the proposed LSTM-s2s with attention can improve inflow forecasting despite its limits in hourly prediction.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.7
• /
• pp.101-108
• /
• 2022

We propose an intrusion detection model that detects denial-of-service(DoS) and distributed reflection denial-of-service(DRDoS) attacks, based on the empirical data of each internet of things(IoT) device by training system and network metrics that can be commonly collected from various IoT devices. First, we collect 37 system and network metrics from each IoT device considering IoT attack scenarios; further, we train them using six types of machine learning models to identify the most effective machine learning models as well as important metrics in detecting and distinguishing IoT attacks. Our experimental results show that the Random Forest model has the best performance with accuracy of over 96%, followed by the K-Nearest Neighbor model and Decision Tree model. Of the 37 metrics, we identified five types of CPU, memory, and network metrics that best imply the characteristics of the attacks in all the experimental scenarios. Furthermore, we found out that packets with higher transmission speeds than larger size packets represent the characteristics of DoS and DRDoS attacks more clearly in IoT networks.

• Journal of Korea Multimedia Society
• /
• v.25 no.10
• /
• pp.1375-1385
• /
• 2022

In this paper, we propose a semi-supervised domain adaptation solution to deal with practical face recognition (FR) scenarios where a single face image for each target identity (to be recognized) is only available in the training phase. Main goal of the proposed method is to reduce the discrepancy between the target and the source domain face images, which ultimately improves FR performances. The proposed method is based on the Domain Adatation network (DAN) using an MMD loss function to reduce the discrepancy between domains. In order to train more effectively, we develop a novel loss function learning strategy in which MMD loss and cross-entropy loss functions are adopted by using different weights according to the progress of each epoch during the learning. The proposed weight adoptation focuses on the training of the source domain in the initial learning phase to learn facial feature information such as eyes, nose, and mouth. After the initial learning is completed, the resulting feature information is used to training a deep network using the target domain images. To evaluate the effectiveness of the proposed method, FR performances were evaluated with pretrained model trained only with CASIA-webface (source images) and fine-tuned model trained only with FERET's gallery (target images) under the same FR scenarios. The experimental results showed that the proposed semi-supervised domain adaptation can be improved by 24.78% compared to the pre-trained model and 28.42% compared to the fine-tuned model. In addition, the proposed method outperformed other state-of-the-arts domain adaptation approaches by 9.41%.