• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.1
• /
• pp.41.1-41.1
• /
• 2021

We generate solar magnetograms and EUV images from Galileo sunspot drawings using a deep learning model based on conditional generative adversarial networks. We train the model using pairs of sunspot drawing from Mount Wilson Observatory (MWO) and their corresponding magnetogram (or UV/EUV images) from 2011 to 2015 except for every June and December by the SDO (Solar Dynamic Observatory) satellite. We evaluate the model by comparing pairs of actual magnetogram (or UV/EUV images) and the corresponding AI-generated one in June and December. Our results show that bipolar structures of the AI-generated magnetograms are consistent with those of the original ones and their unsigned magnetic fluxes (or intensities) are well consistent with those of the original ones. Applying this model to the Galileo sunspot drawings in 1612, we generate HMI-like magnetograms and AIA-like EUV images of the sunspots. We hope that the EUV intensities can be used for estimating solar EUV irradiance at long-term historical times.

• Journal of Korean Society for Atmospheric Environment
• /
• v.10 no.4
• /
• pp.224-232
• /
• 1994

A new neural network model has been developed to predict short-term air pollution concentration. In addition, a multiple regression model widely used in statistical analysis was tested. These models were applied for prediction of daily maximum ozone concentration in Seoul during the summer season of 1991. The time periods between May and September 1989 and 1990 were utilized to train set of learning patterns in neural network model, and to estimate multiple regression model. To evaluate the results of the different models, several Performance indices were used. The results indicated that the multiple regression model tended to underpredict the daily maximum ozone concentration with small r$^{2}$(0.38). Also, large errors were found in this model; 21.1 ppb for RMSE, 0.324 for NMSE, and -0.164 for MRE. On the other hand, the results obtained from the neural network model were very promising. Thus, we can know that this model has a prominent efficiency in the adaptive control for the non-linear multi- variable systems such as photochemical oxidants. Also, when the recent new information was added in the neural network model, prediction accuracy was increased. From the new model, the values of RMSE, NMSE and r$^{2}$ were 13.2ppb, 0.089, 0.003 and 0.55 respectively.

• Journal of the Korean Society for Railway
• /
• v.13 no.2
• /
• pp.166-172
• /
• 2010

Currently, vector control is used for speed control of trains because induction motor has high performance is installed in Electric railroad systems. Also, control of the induction motor is possible through various methods by developing inverters and control theory. Presently, rolling stocks which use the induction motor are possible to brake trains by using AC motor. Therefore model of motor block and induction motor is needed to adapt various methods. There is Variable Voltage Variable Frequency (VVVF) as the control method of the induction motor. The torque and speed is controlled in the VVVF. The propulsion system model in the electric railroad has many sub-systems. So, the analysis of performance of the speed control is very complex. In this paper, simulation models are suggested by using Matlab/Simulink in the speed control characteristic. On the basis of the simulation models, the response to disturbance input is analyzed about the load. Also, the current, speed and flux control model are proposed to analyze the speed control characteristic in the train propulsion system.

• Journal of Advanced Marine Engineering and Technology
• /
• v.35 no.2
• /
• pp.216-223
• /
• 2011

TVC is a kind of ejector which entrains low pressure working fluid by using the high pressure working fluid. While most papers relating with ejectors treat the working fluid as an ideal gas for convenience, the fluid doesn't behave as the ideal gas when phase change occurs. In this study, numerical analysis is conducted by applying Redlich-Kwong equation of state instead of ideal gas equation of state. Two turbulent models are compared for the better prediction and SST k-${\omega}$ model is preferred rather than realizable k-${\epsilon}$ model by comparison. Energy loss at the diffuser inlet and throat using the real gas equation of state is relatively greater than that using ideal gas law. For the real gas case, pressure increase due to shock train at the diffuser outlet is relatively smaller than the ideal gas case, but both cases have the same pressure increase due to a pseudo shock.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.3
• /
• pp.199-207
• /
• 2021

This study aimed to improve the performance of the KTX (Korea Train Express) brake system. To develop a braking disc-hub for the high-speed rail, the model performance was analyzed by finite element analysis, and the analysis results were verified using the braking test results. In addition, heat transfer analysis, thermal stress analysis, natural frequency analysis, and static analysis were conducted to examine the mechanical performance of the braking system. By deriving the design factors and conducting parametric analyses according to the shape of the hub, this study derived the optimal specifications that could improve heat dissipation and reduce weight. The cooling efficiency and structural performance of the optimization model were improved during braking compared to the existing model. It is expected that the design verification will be carried out through analyses of the optimal specifications so that it can be used in the development of brakes in railway vehicles and motor vehicles.

• Journal of the Korean Institute of Electrical and Electronic Material Engineers
• /
• v.35 no.2
• /
• pp.134-142
• /
• 2022

As industry and technology go through advancement, it is hard to search new materials which satisfy various standards through conventional trial-and-error based research methods. Crystal Graph Convolutional Neural Network(CGCNN) is a neural network which uses material's features as train data, and predicts the material properties(formation energy, bandgap, etc.) much faster than first-principles calculation. This report introduces how to train the CGCNN model which predicts the formation energy using open database. It is anticipated that with a simple programming skill, readers could construct a model using their data and purpose. Developing machine learning model for materials science is going to help researchers who should explore large chemical and structural space to discover materials efficiently.

• Computers and Concrete
• /
• v.28 no.6
• /
• pp.533-547
• /
• 2021

Calculating the shear capacity of slender reinforced concrete beams without shear reinforcement was the subject of numerous studies, where the eternal problem of developing a single relationship that will be able to predict the expected shear capacity is still present. Using experimental results to extrapolate formulae was so far the main approach for solving this problem, whereas in the last two decades different research studies attempted to use artificial intelligence algorithms and available data sets of experimentally tested beams to develop new models that would demonstrate improved prediction capabilities. Given the limited number of available experimental databases, these studies were numerically restrained, unable to holistically address this problem. In this manuscript, a new approach is proposed where a numerically generated database is used to train machine-learning algorithms and develop an improved model for predicting the shear capacity of slender concrete beams reinforced only with longitudinal rebars. Finally, the proposed predictive model was validated through the use of an available ACI database that was developed by using experimental results on physical reinforced concrete beam specimens without shear and compressive reinforcement. For the first time, a numerically generated database was used to train a model for computing the shear capacity of slender concrete beams without stirrups and was found to have improved predictive abilities compared to the corresponding ACI equations. According to the analysis performed in this research work, it is deemed necessary to further enrich the current numerically generated database with additional data to further improve the dataset used for training and extrapolation. Finally, future research work foresees the study of beams with stirrups and deep beams for the development of improved predictive models.

• Journal of The Korean Association of Information Education
• /
• v.25 no.6
• /
• pp.1015-1024
• /
• 2021

There are various educational programming environments in which students can train artificial intelligence (AI) using block-based programming languages, such as Entry, Machine Learning for Kids, and Teachable Machine. However, these programming environments are designed so that students can train AI through a separate menu, and then use the trained model in the code editor. These approaches have the advantage that students can check the training process more intuitively, but there is also the disadvantage that both the training menu and the code editor must be used. In this paper, we present a novel artificial intelligence block that can perform both AI training and programming in the code editor. While this AI block is presented as a Scratch block, the training process is performed through a Python server. We describe the blocks in detail through the process of training a model to classify a blue pen and a red pen, and a model to classify a dental mask and a KF94 mask. Also, we experimentally show that our approach is not significantly different from Teachable Machine in terms of performance.

• Journal of the Korean Society of Radiology
• /
• v.17 no.5
• /
• pp.709-715
• /
• 2023

In this study, the performance evaluation of image quality for noise reduction was implemented using the U-net deep learning architecture in computed tomography (CT) images. In order to generate input data, the Gaussian noise was applied to ground truth (GT) data, and datasets were consisted of 8:1:1 ratio of train, validation, and test sets among 1300 CT images. The Adagrad, Adam, and AdamW were used as optimizer function, and 10, 50 and 100 times for number of epochs were applied. In addition, learning rates of 0.01, 0.001, and 0.0001 were applied using the U-net deep learning model to compare the output image quality. To analyze the quantitative values, the peak signal to noise ratio (PSNR) and coefficient of variation (COV) were calculated. Based on the results, deep learning model was useful for noise reduction. We suggested that optimized hyper parameters for noise reduction in CT images were AdamW optimizer function, 100 times number of epochs and 0.0001 learning rates.

• The Journal of Korean Medicine
• /
• v.44 no.4
• /
• pp.136-149
• /
• 2023

Objectives: Facial diagnosis is an important part of clinical diagnosis in traditional East Asian Medicine. In this paper, we proposed a model to quantitatively classify cold sensitivity using a fully automated facial image analysis system. Methods: We investigated cold sensitivity in 452 subjects. Cold sensitivity was determined using a questionnaire and the Cold Pattern Score (CPS) was used for analysis. Subjects with a CPS score below the first quartile (low CPS group) belonged to the cold non-sensitivity group, and subjects with a CPS score above the third quartile (high CPS group) belonged to the cold sensitivity group. After splitting the facial images into train/validation/test sets, the train and validation set were input into a convolutional neural network to learn the model, and then the classification accuracy was calculated for the test set. Results: The classification accuracy of the low CPS group and high CPS group using facial images in all subjects was 76.17%. The classification accuracy by sex was 69.91% for female and 62.86% for male. It is presumed that the deep learning model used facial color or facial shape to classify the low CPS group and the high CPS group, but it is difficult to specifically determine which feature was more important. Conclusions: The experimental results of this study showed that the low CPS group and the high CPS group can be classified with a modest level of accuracy using only facial images. There was a need to develop more advanced models to increase classification accuracy.