• Journal of Korea Water Resources Association
• /
• v.41 no.2
• /
• pp.149-162
• /
• 2008

The Soil and Water Assessment Tool (SWAT) developed by the USDA-Agricultural Research Service for the prediction of land management impact on water, sediment, and agricultural chemical yields in a large-scale basin was applied to Daecheong Reservoir basin to estimate the amount of soil losses from different land uses. The research outcomes provide important indications for reservoir managers and policy makers to search alternative watershed management practices for the mitigation of reservoir turbidity flow problems. After calibrations of key model parameters, SWAT showed fairly good performance by adequately simulating observed annual runoff components and replicating the monthly flow regimes in the basin. The specific soil losses from agricultural farm field, forest, urban area, and paddy field were 33.1, $2.3{\sim}5.4$ depending on the tree types, 1.0, and 0.1 tons/ha/yr, respectively in 2004. It was noticed that about 55.3% of the total annual soil loss is caused by agricultural activities although agricultural land occupies only 10% in the basin. Although the soil erosion assessment approach adopted in this study has some extent of uncertainties due to the lack of detailed information on crop types and management activities, the results at least imply that soil erosion control practices for the vulnerable agricultural farm lands can be one of the most effective alternatives to reduce the impact of turbidity flow in the river basin system.

• Ecology and Resilient Infrastructure
• /
• v.7 no.2
• /
• pp.114-125
• /
• 2020

Despite the various artificial neural networks that have been developed, most of the discharge models in previous studies have been developed using deep neural networks. This study aimed to develop a discharge model using a convolution neural network (CNN), which was used to solve classification problems. Furthermore, the applicability of CNN was evaluated. The photographs (pictures or images) for input data to CNN could not clearly show the characteristics of the study area as well as precipitation. Hence, the model employed in this study had to use numerical images. To solve the problem, the CN of NRCS was used to generate images as input data for the model. The generated images showed a good possibility of applicability as input data. Moreover, a new application of CN, which had been used only for discharge prediction, was proposed in this study. As a result of CNN training, the model was trained and generalized stably. Comparison between the actual and predicted values had an R² of 0.79, which was relatively high. The model showed good performance in terms of the Pearson correlation coefficient (0.84), the Nash-Sutcliffe efficiency (NSE) (0.63), and the root mean square error (24.54 ㎥/s).

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.42 no.4
• /
• pp.326-335
• /
• 2014

Aerodynamic characteristics of the small two-stage turbo blower were investigated using commercial CFD tool(ANSYS CFX Ver. 14.5) in this paper. Turbo blower, which is a centrifugal type of turbomachinery, is used in various industries. It is used for application that required high static pressure rising at relatively small volumetric flow rate. In order to understand the mechanism of static pressure rising, the aerodynamic characteristics of the small two-stage turbo blower are analyzed at high rotating speed in this study. The k-${\omega}$ SST turbulence model, which is good at prediction of adverse pressure gradient flows, was applied. The CFD results of the turbo blower are validated by performance test. The static pressure rising of the turbo blower is nonlinearly increased over the first stage and the second stage. The secondary flow occurred at guide vanes, between the casing and the first impeller shroud, and the bottom of the impeller disk. As a result, It is required that whole fluid area is analyzed to predict aerodynamic characteristics of small two-stage turbo blower. and the result should be selected with considering for error from experiment and CFD.

• Journal of Energy Engineering
• /
• v.24 no.4
• /
• pp.183-191
• /
• 2015

Studies on the environment-friendly and permanent low energy saving measure are and will continue to be an eternal challenge. However, the demand is high for the technologies that can save energy significantly in everyday life that produce tangible benefits for the users by applying saving factors and that anyone can access easily when it comes the related procedure. Government policies related to the improvement of energy effect in the existing building structure are characterized by complex procedure. Moreover, cost required and reliability issue emerge when request is made to an expert. Accordingly, this study seeks to present energy improvement plan that can be utilized simply and conveniently at any place, any time by enabling customized design according to individual taste by enabling energy change amount and grade prediction when the users select only the part that they want to replace by using a simple program.

• Journal of Internet Computing and Services
• /
• v.21 no.5
• /
• pp.1-7
• /
• 2020

Deep neural networks(DNN), which are used as approximation functions in reinforcement learning (RN), theoretically can be attributed to realistic results. In empirical benchmark works, time difference learning (TD) shows better results than Monte-Carlo learning (MC). However, among some previous works show that MC is better than TD when the reward is very rare or delayed. Also, another recent research shows when the information observed by the agent from the environment is partial on complex control works, it indicates that the MC prediction is superior to the TD-based methods. Most of these environments can be regarded as 5-step Q-learning or 20-step Q-learning, where the experiment continues without long roll-outs for alleviating reduce performance degradation. In other words, for networks with a noise, a representative network that is regardless of the controlled roll-outs, it is better to learn MC, which is robust to noisy rewards than TD, or almost identical to MC. These studies provide a break with that TD is better than MC. These recent research results show that the way combining MC and TD is better than the theoretical one. Therefore, in this study, based on the results shown in previous studies, we attempt to exploit a random balance with a mixture of TD and MC in RL without any complicated formulas by rewards used in those studies do. Compared to the DQN using the MC and TD random mixture and the well-known DQN using only the TD-based learning, we demonstrate that a well-performed TD learning are also granted special favor of the mixture of TD and MC through an experiments in OpenAI Gym.

• Journal of Embryo Transfer
• /
• v.25 no.4
• /
• pp.263-266
• /
• 2010

Prediction of semen's fertilizing ability used in artificial insemination (AI) is one of very important factors on pig reproductive performance. In vitro fertilization (IVF) has been used for indirect evaluation of sperm's fertilizing ability and it has been showed as highly correlated index. In swine industry, increasing interest in preservation of boar semen raises questions on the sperm motility from semen used in commercial AI centers. Mitochondria in sperm mid-piece generate the energy to support motility and could be an explanation of impaired fertility. Objective of this study was to suggest usable sperm motility to farms in measuring the effect of sperm motility and sperm abnormality on in vitro production of embryo in which sperm's fertilizing ability can be determined indirectly. Semen samples were provided from local AI center and used within 3 days after collection. Semen samples were divided by 4 different motile groups (>70%; 61~70%; 51~60%; <50%) using CASA (computer-assisted sperm analysis) on the days of IVF. Developmental rate to the blastocyst stage from over 61% motile sperm group showed significantly higher rate than below 60% motile sperm group ($16.5{\pm}0.7{\sim}18.4{\pm}0.8%$ vs $6.3{\pm}0.8{\sim}11.5{\pm}0.7%$, p<0.05). In experiment to determine the relationship between sperm motility and viability and abnormality, over 61% motile sperm groups showed significantly higher viability rate compared to below 60% motile sperm groups ($84.8{\pm}4.0{\sim}88.1{\pm}4.0%$ vs $69.1{\pm}4.0{\sim}74.2{\pm}4.0%$, p<0.05). On the other hand, morphological sperm abnormality showed significantly higher in over 70% motile sperm group ($10.2{\pm}2.2$ vs $16.0{\pm}2.2{\sim}21.0{\pm}2.2%$, p<0.05). In experiment to find the correlation between sperm motility of 4 different motile groups and amount of mitochondria, lower motility group also showed lower level of mitochondria (p<0.05). The mitochondria parameter used in this study showed another possibility to differentiate the sperm motility. Taken together, because below 60% motile semen used in AI reduce the fertility, AI centers should provide the over 60% motile sperm to the farms at the time of AI.

• Clinical and Experimental Reproductive Medicine
• /
• v.43 no.2
• /
• pp.112-118
• /
• 2016

Objective: Ovarian reserve tests are commonly used to predict ovarian response in infertile patients undergoing ovarian stimulation. Although serum markers such as basal follicle-stimulating hormone (FSH) or random $anti-M{\ddot{u}}llerian$ hormone (AMH) level and ultrasonographic markers (antral follicle count, AFC) are good predictors, no single test has proven to be the best predictor. In this study, we developed appropriate equations and novel nomograms to predict the number of oocytes that will be retrieved using patients' age, serum levels of basal FSH and AMH, and AFC. Methods: We analyzed a database containing clinical and laboratory information of 141 stimulated in vitro fertilization (IVF) cycles performed at a university-based hospital between September 2009 and December 2013. We used generalized linear models for prediction of the number of oocytes. Results: Age, basal serum FSH level, serum AMH level, and AFC were significantly related to the number of oocytes retrieved according to the univariate and multivariate analyses. The equations that predicted the number of oocytes retrieved (log scale) were as follows: model (1) $3.21-0.036{\times}(age)+0.089{\times}(AMH)$, model (2) $3.422-0.03{\times}(age)-0.049{\times}(FSH)+0.08{\times}(AMH)$, model (3) $2.32-0.017{\times}(age)+0.039{\times}(AMH)+0.03{\times}(AFC)$, model (4) $2.584-0.015{\times}(age)-0.035{\times}(FSH)+0.038{\times}(AMH)+0.026{\times}(AFC)$. model 4 showed the best performance. On the basis of these variables, we developed nomograms to predict the number of oocytes that can be retrieved. Conclusion: Our nomograms helped predict the number of oocytes retrieved in stimulated IVF cycles.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.22 no.7
• /
• pp.782-790
• /
• 2016

When a fire occurs on a ship that has mechanical ventilation facilities, the air supply and exhaust systems directly effect smoke diffusion. And there is a high possibility that occupant's visibility will be harmed because of smoke. In this study, the effects and risks of air supply and exhaust systems with regard to smoke diffusion given a shipboard fire analyzed with a Fire Dynamic Simulator(FDS). Suggested measures are also provided for using air supply and exhaust systems more efficiently. The results showed that, when air supply and exhaust systems were both working at the time of a fire, rather than stopping these systems as previously encouraged, continuing to operate both was an effective measure to gain evacuation time. When a fire occurred and the exhaust system was operating, also starting the air supply system near the origin of the fire was another effective approach to gain evacuation time. However, when only the air supply system was operating and a fire occurred, the air supply system accelerated smoke diffusion, so it was necessary to stop the air supply system to detect smoke diffusion as much as possible.

• Current Optics and Photonics
• /
• v.1 no.2
• /
• pp.107-112
• /
• 2017

An adaptive optics system can be simulated or analyzed to predict its closed-loop performance. However, this type of prediction based on various assumptions can occasionally produce outcomes which are far from actual experience. Thus, every adaptive optics system is desired to be tested in a closed loop on an optical test bench before its application to a telescope. In the close-loop test bench, we need an atmospheric simulator that simulates atmospheric disturbances, mostly in phase, in terms of spatial and temporal behavior. We report the development of an atmospheric turbulence simulator consisting of two point sources, a commercially available deformable mirror with a $12{\times}12$ actuator array, and two random phase plates. The simulator generates an atmospherically distorted single or binary star with varying stellar magnitudes and angular separations. We conduct a simulation of a binary star by optically combining two point sources mounted on independent precision stages. The light intensity of each source (an LED with a pin hole) is adjustable to the corresponding stellar magnitude, while its angular separation is precisely adjusted by moving the corresponding stage. First, the atmospheric phase disturbance at a single instance, i.e., a phase screen, is generated via a computer simulation based on the thin-layer Kolmogorov atmospheric model and its temporal evolution is predicted based on the frozen flow hypothesis. The deformable mirror is then continuously best-fitted to the time-sequenced phase screens based on the least square method. Similarly, we also implement another simulation by rotating two random phase plates which were manufactured to have atmospheric-disturbance-like residual aberrations. This later method is limited in its ability to simulate atmospheric disturbances, but it is easy and inexpensive to implement. With these two methods, individually or in unison, we can simulate typical atmospheric disturbances observed at the Bohyun Observatory in South Korea, which corresponds to an area from 7 to 15 cm with regard to the Fried parameter at a telescope pupil plane of 500 nm.

• Radiation Oncology Journal
• /
• v.36 no.2
• /
• pp.139-146
• /
• 2018

Purpose: IBTR! 2.0 nomogram is web-based nomogram that predicts ipsilateral breast tumor recurrence (IBTR). We aimed to validate the IBTR! 2.0 using an external data set. Materials and Methods: The cohort consisted of 2,206 patients, who received breast conserving surgery and radiation therapy from 1992 to 2012 at our institution, where wide surgical excision is been routinely performed. Discrimination and calibration were used for assessing model performance. Patients with predicted 10-year IBTR risk based on an IBTR! 2.0 nomogram score of <3%, 3%-5%, 5%-10%, and >10% were assigned to groups 1, 2, 3, and 4, respectively. We also plotted calibration values to observe the actual IBTR rate against the nomogram-derived 10-year IBTR probabilities. Results: The median follow-up period was 73 months (range, 6 to 277 months). The area under the receiver operating characteristic curve was 0.607, showing poor accordance between the estimated and observed recurrence rate. Calibration plot confirmed that the IBTR! 2.0 nomogram predicted the 10-year IBTR risk higher than the observed IBTR rates in all groups. High discrepancies between nomogram IBTR predictions and observed IBTR rates were observed in overall risk groups. Compared with the original development dataset, our patients had fewer high grade tumors, less margin positivity, and less lymphovascular invasion, and more use of modern systemic therapies. Conclusions: IBTR! 2.0 nomogram seems to have the moderate discriminative ability with a tendency to over-estimating risk rate. Continued efforts are needed to ensure external applicability of published nomograms by validating the program using an external patient population.