• Journal of Navigation and Port Research
• /
• v.47 no.4
• /
• pp.231-238
• /
• 2023

The position of a light buoy is always flexible due to the influence of external forces such as tides and wind. The position can be checked through AIS (Automatic Identification System) or RTU (Remote Terminal Unit) for AtoN. As a result of analyzing the position data for the last five years (2017-2021) of a light buoy, the average position error was 15.4%. It is necessary to detect position error data and obtain refined position data to prevent navigation safety accidents and management. This study aimed to detect position error data and obtain refined position data by DBSCAN Clustering position data obtained through AIS or RTU for AtoN. For this purpose, 21 position data of Gunsan Port No. 1 light buoy where RTU was installed among western waters with the most position errors were DBSCAN clustered using Python library. The minPts required for DBSCAN Clustering applied the value commonly used for two-dimensional data. Epsilon was calculated and its value was applied using the k-NN (nearest neighbor) algorithm. As a result of DBSCAN Clustering, position error data that did not satisfy minPts and epsilon were detected and refined position data were acquired. This study can be used as asic data for obtaining reliable position data of a light buoy installed with AIS or RTU for AtoN. It is expected to be of great help in preventing navigation safety accidents.

• Journal of the Korean Society of Radiology
• /
• v.17 no.4
• /
• pp.557-564
• /
• 2023

In this study, 3D printing technology was used to compensate for the shortcomings of the use of lead, which has proven to have excellent shielding performance, and to control unnecessary human exposure. 3D printers can implement three-dimensional shapes and can immediately apply individual ideas, which has great advantages in maintaining technology supplementation while reducing the cost and duration of prototyping. Among the various special 3D printers, the FDM method was adopted, and the filament used for output was manufactured using a research extruder by mixing two materials, PLA (Poly-Lactic-Acid) and tungsten. The purpose was to verify the validity through dose evaluation and to provide basic information on the production of chapezones of various materials. The mixed filament was implemented as a morphological shield. Filaments made of a research extruder by mixing PLA and tungsten were divided into 10 %, 20 %, 30 %, 40 %, and 50 % according to the tungsten content ratio. Through the process of 3D Modeling, STL File storage, G-code generation, and output, 10 cm × 10 cm × 0.5 cm was manufactured, respectively, and dose and shielding ability were evaluated under the conditions of tube voltages of 60 kVp, 80 kVp, 100 kVp, 120 kVp, and tube currents of 20 mAs and 40 mAs.

• Journal of Broadcast Engineering
• /
• v.28 no.3
• /
• pp.293-301
• /
• 2023

Unstandardized medical data collection and management are still being conducted manually, and studies are being conducted to classify CT data using deep learning to solve this problem. However, most studies are developing models based only on the axial plane, which is a basic CT slice. Because CT images depict only human structures unlike general images, reconstructing CT scans alone can provide richer physical features. This study seeks to find ways to achieve higher performance through various methods of converting CT scan to 2D as well as axial planes. The training used 1042 CT scans from five body parts and collected 179 test sets and 448 with external datasets for model evaluation. To develop a deep learning model, we used InceptionResNetV2 pre-trained with ImageNet as a backbone and re-trained the entire layer of the model. As a result of the experiment, the reconstruction data model achieved 99.33% in body part classification, 1.12% higher than the axial model, and the axial model was higher only in brain and neck in contrast classification. In conclusion, it was possible to achieve more accurate performance when learning with data that shows better anatomical features than when trained with axial slice alone.

• Korean Chemical Engineering Research
• /
• v.61 no.1
• /
• pp.123-141
• /
• 2023

A sufficient amount of data with quality is needed for training artificial neural networks (ANNs). However, developing ANN models with a small amount of data often appears in engineering fields. This paper presented an ANN model to improve prediction performance of the ammonia emission amount with 83 data. The ammonia emission rate included eleven inputs and two outputs (maximum ammonia loss, N_max and time to reach half of N_max, K_m). Categorical input variables were transformed into multi-dimensional equal-distance variables, and 13 data were added into 66 training data using a generative adversarial network. Hyperparameters (number of layers, number of neurons, and activation function) of ANN were optimized using Gaussian process. Using 17 test data, the previous ANN model (Lim et al., 2007) showed the mean absolute error (MAE) of K_m and N_max to 0.0668 and 0.1860, respectively. The present ANN outperformed the previous model, reducing MAE by 38% and 56%.

• Nuclear Engineering and Technology
• /
• v.54 no.12
• /
• pp.4571-4584
• /
• 2022

The analysis of rapid flow transients in Reactor Coolant Pumps (RCP) is essential for a reactor safety study. An accurate and precise analysis of the RCP coastdown is necessary for the reactor design. The coastdown of RCP affects the coolant temperature and the colloidal crud in the primary coolant. A realistic and kinetic model has been used to investigate the behavior of activated colloidal crud in the primary coolant and steam generator that solves the pump speed analytically. The analytic solution of the non-dimensional flow rate has been determined by the energy ratio β. The kinetic energy of the coolant fluid and the kinetic energy stored in the rotating parts of a pump are two essential parameters in the form of β. Under normal operation, the pump's speed and moment of inertia are constant. However, in a coastdown situation, kinetic damping in the interval has been implemented. A dynamic model ACCP-SMART has been developed for System Integrated Modular and Advanced Reactor (SMART) to investigate the corrosion due to activated colloidal crud. The Fickian diffusion model has been implemented as the reference corrosion model for the constituent component of the primary loop of the SMART reactor. The activated colloidal crud activity in the primary coolant and steam generator of the SMART reactor has been studied for different equilibrium corrosion rates, linear increase in corrosion rate, and dynamic RCP coastdown situation energy ratio b. The coolant specific activity of SMART reactor equilibrium corrosion (4.0 mg s^-1) has been found 9.63×10^-3 µCi cm^-3, 3.53×10^-3 µC cm^-3, 2.39×10^-2 µC cm^-3, 8.10×10^-3 µC cm^-3, 6.77× 10^-3 µC cm^-3, 4.95×10^-4 µC cm^-3, 1.19×10^-3 µC cm^-3, and 7.87×10^-4 µC cm^-3 for ²⁴Na, ⁵⁴Mn, ⁵⁶Mn, ⁵⁹Fe, ⁵⁸Co, ⁶⁰Co, ⁹⁹Mo, and ⁵¹Cr which are 14.95%, 5.48%, 37.08%, 12.57%, 10.51%, 0.77%, 18.50%, and 0.12% respectively. For linear and exponential coastdown with a constant corrosion rate, the total coolant and steam generator activity approaches a higher saturation value than the normal values. The coolant and steam generator activity changes considerably with kinetic corrosion rate, equilibrium corrosion, growth of corrosion rate (ΔC/Δt), and RCP coastdown situations. The effect of the RCP coastdown on the specific activity of the steam generators is smeared by linearly rising corrosion rates, equilibrium corrosion, and rapid coasting down of the RCP. However, the time taken to reach the saturation activity is also influenced by the slope of corrosion rate, coastdown situation, equilibrium corrosion rate, and energy ratio β.

• Journal of Korean Society of Forest Science
• /
• v.112 no.2
• /
• pp.195-208
• /
• 2023

This study aimed to use three-dimensional point cloud data (PCD) obtained from Terrestrial Laser Scanning (TLS) and Mobile Laser Scanning (MLS) to evaluate a deep learning-based species classification model for two tree species: Pinus koraiensis and Larix kaempferi. Sixteen models were constructed based on the three conditions: LiDAR platform (TLS and MLS), down-sampling intensity (1024, 2048, 4096, 8192), and deep learning model (PointNet, PointNet++). According to the classification accuracy evaluation, the highest kappa coefficients were 93.7% for TLS and 96.9% for MLS when applied to PCD data from the PointNet++ model, with down-sampling intensities of 8192 and 2048, respectively. Furthermore, PointNet++ was consistently more accurate than PointNet in all scenarios sharing the same platform and down-sampling intensity. Misclassification occurred among individuals of different species with structurally similar characteristics, among individual trees that exhibited eccentric growth due to their location on slopes or around trails, and among some individual trees in which the crown was vertically divided during tree segmentation.

• Journal of the Korean Geotechnical Society
• /
• v.24 no.1
• /
• pp.111-118
• /
• 2008

In the study a 2 dimensional model test was executed to grasp the effect of the taking load of equipments on the ground when improving a soft ground like dredging reclaimed ground. The static load and the dynamic load in the consolidated model ground was $0.02kg/cm^2,\;0.03kg/cm^2\;and\;0.04kg/cm^2$ respectively. After consolidating far two months by consolidation load of $0.02kg/cm^2,\;0.03kg/cm^2\;and\;0.04kg/cm^2$ respectively, the ultimate bearing capacity was $0.16kg/cm^2,\;0.19kg/cm^2,\;0.24kg/cm^2$ respectively. And the energy price of dynamic load test at the same point as the settlement of static load test indicated $E=336{\sim}945kg{\cdot}cm,\;E=252{\sim}780kg{\cdot}cm\;and\;E=323{\sim}727kg{\cdot}cm$ for each consolidation load. When the static load and the dynamic load operated at the same ground condition, the heaving quantity was bigger in the case of the dynamic load than in the case of the static load, and the horizontal displacement quantity the in the case of dynamic load was exhibited very deficiently compared to the quantity in the case of static load test.

• Journal of the Korean Geotechnical Society
• /
• v.23 no.12
• /
• pp.83-94
• /
• 2007

Piled raft foundations have been highlighted as an economical design concept of pile foundations in recent years. However, piled raft foundations have not been widely used in Korea due to the difficulty in estimating the complex interaction effects among rafts, piles and soils. The authors developed an effective numerical program to analyze the behavior of piled raft foundations for practical design purposes and presented it briefly in this paper. The developed numerical program simulates the raft as a flexible plate consisting of finite elements with eight nodes and the raft is supported by a series of elastic springs representing subsoils and piles. This study imported another model to simulate pile groups considering non-linear behavior and interaction effects. The apparent stiffnesses of the soils and piles were estimated by iterative calculations to satisfy the compatibility between those two components and the behavior of piled raft foundations can be predicted using these stiffnesses. For the verification of the program, the analysis results about some example problems were compared with those of rigorous three dimensional finite element analysis and other approximate analysis methods. It was found that the program can analyze non-linear behaviors and interaction effects efficiently in multi-layered soils and has sufficient capabilities for application to practical analysis and design of piled raft foundations.

• Journal of the Korean Geotechnical Society
• /
• v.24 no.12
• /
• pp.113-120
• /
• 2008

One-dimensional site response analysis is widely used to simulate the seismic site effects. The equivalent linear analysis, which is the most widely used type of site response analysis, is essentially a linear method. The method applies constant shear modulus and damping throughout the frequency range of the input motion, ignoring the dependence of the soil response on the loading frequency. A new type of equivalent linear analysis method that can simulate the frequency dependence of the soil behavior via frequency-strain curve was developed. Various forms of frequency-strain curves were proposed, and all curves were asserted to increase the accuracy of the solution. However, its validity has not been extensively proven and the effect of the shape of the frequency-strain curve is not known. This paper used two previously proposed frequency-strain curves and three additional curves developed in this study to evaluate the accuracy of the frequency-dependent equivalent linear method and the influence of the shape of the frequency-strain curves. In the evaluation, six recordings from three case histories were used. The results of the case study indicated that the shape of the frequency-strain curve has a dominant influence on the calculated response, and that the frequency dependent analysis can enhance the accuracy of the solution. However, a curve that results in the best match for all case histories did not exist and the optimum curve varied for each case. Since the optimum frequency-strain curve can not be defined, it is recommended that a suite of curves be used in the analysis.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.23 no.3
• /
• pp.145-151
• /
• 2023

3D-CNN is one of the deep learning techniques for learning time series data. Such three-dimensional learning can generate many parameters, so that high-performance machine learning is required or can have a large impact on the learning rate. When learning dynamic hand-gestures in spatiotemporal domain, it is necessary for the improvement of the efficiency of dynamic hand-gesture learning with 3D-CNN to find the optimal conditions of input video data by analyzing the learning accuracy according to the spatiotemporal change of input video data without structural change of the 3D-CNN model. First, the time ratio between dynamic hand-gesture actions is adjusted by setting the learning interval of image frames in the dynamic hand-gesture video data. Second, through 2D cross-correlation analysis between classes, similarity between image frames of input video data is measured and normalized to obtain an average value between frames and analyze learning accuracy. Based on this analysis, this work proposed two methods to effectively select input video data for 3D-CNN deep learning of dynamic hand-gestures. Experimental results showed that the learning interval of image data frames and the similarity of image frames between classes can affect the accuracy of the learning model.