• Journal of Korea Water Resources Association
• /
• v.53 no.4
• /
• pp.273-283
• /
• 2020

Quantitative forecasting of groundwater levels for the assessment of groundwater variation and vulnerability is very important. To achieve this purpose, various time series analysis and machine learning techniques have been used. In this study, we developed a prediction model based on LSTM (Long short term memory), one of the artificial neural network (ANN) algorithms, for predicting the daily groundwater level of 11 groundwater wells in Hankyung-myeon, Jeju Island. In general, the groundwater level in Jeju Island is highly autocorrelated with tides and reflected the effects of precipitation. In order to construct an input and output variables based on the characteristics of addressing data, the precipitation data of the corresponding period was added to the groundwater level data. The LSTM neural network was trained using the initial 365-day data showing the four seasons and the remaining data were used for verification to evaluate the fitness of the predictive model. The model was developed using Keras, a Python-based deep learning framework, and the NVIDIA CUDA architecture was implemented to enhance the learning speed. As a result of learning and verifying the groundwater level variation using the LSTM neural network, the coefficient of determination (R²) was 0.98 on average, indicating that the predictive model developed was very accurate.

• The Korea Journal of Herbology
• /
• v.21 no.2
• /
• pp.109-119
• /
• 2006

Objectives : This study was performed to approach of properties-flavours theory and the study of morphological standard in CORNI FRUCTUS. Methods : The properties-flavours theory were attempted with bibliographic method and various climate-information. The external-internal morphological standards were determined by using stereoscope and butanol series. Results and conclusion : 1. The significant analysis for the corni fructus as the properties-flavours theory, is a follow-up survey of the effect written in the official oriental medicine book. 2. The other analytic methods according to the gathering time, processing or storing system or especially nature or growth environment(altitude, temperature,. the mean moisture, and the agronomical survey) by way of the explanation of properties-flavours theory appears so many insufficiency at the many cases. 3. The external and internal characteristics according to the shape of original plant or herbal states, entirely correspond to all the literature on this subject. 4. According to the place of production, there are some different points. (l) In the external shapes, it is reasonable that if we make the basic standard of classification by the sharpness of color, the existence of gloss and the size of products. In the case of the Korean products, it shows bright and glossy color totally, but in the case of North Korea products, it shows black and lackluster. And in the case of the foreign products, it has softer body than Korea one and shows white color on the surface. (2) In the internal shapes, the Korea products has a thicker cuticle layer than the foreign one, and has fewer ca-oxalate in the parenchyma cell.

• Journal of Digital Convergence
• /
• v.18 no.11
• /
• pp.259-266
• /
• 2020

Many developments have been steadily carried out by researchers with applying knowledge-based expert system or machine learning algorithms to the financial field. In particular, it is now common to perform knowledge based system trading in using stock prices. Recently, deep learning technologies have been applied to real fields of stock trading marketplace as GPU performance and large scaled data have been supported enough. Especially, LSTM has been tried to apply to stock price prediction because of its compatibility for time series data. In this paper, we implement stock price prediction using LSTM. In modeling of LSTM, we propose a fitness combination of model parameters and activation functions for best performance. Specifically, we propose suitable selection methods of initializers of weights and bias, regularizers to avoid over-fitting, activation functions and optimization methods. We also compare model performances according to the different selections of the above important modeling considering factors on the real-world stock price data of global major companies. Finally, our experimental work brings a fitness method of applying LSTM model to stock price prediction.

• Tunnel and Underground Space
• /
• v.12 no.3
• /
• pp.142-151
• /
• 2002

Claesson(2001)'s analytical solution, and two numerical models with Dirichlet and Neuman interior boundary condition respectively were investigated to estimate the transient temperature distribution with distances from the Taejon underground food cold storage pilot cavern. Claesson's solution, which is based on constant temperature boundary condition at the rock wall during a temperature decline step, showed relatively good agreement with temperature measurements in the rock mass in order of average error difference, 0.89$\^{C}$ without any adjustments on laboratory thermal properties to represent the rock mass. For the numerical model with heat flux through the rock wall, a boundary condition setting technique was newly proposed to overcome the difficulty of prescribing variable convective heat tranfer coefficient and far-field air temperature inside the cavern as they may be certainly changed according to the cooling-down time. The results showed also good agreement with measurements in order of average error difference, 1.58$\^{C}$, and were compared to those of the numerical model with fixed temperature at the rock wall. Finally, the most proper procedure to precisely predict the temperature profile around a cavern was proposed as a series of analysis steps including an analytical exact solution and numerical models.

• Publications of The Korean Astronomical Society
• /
• v.30 no.3
• /
• pp.821-832
• /
• 2015

As a part of the short-period variability survey (SPVS) at Bohyunsan Optical Astronomy Observatory, we obtained time-series BV CCD images in the region of the open cluster NGC 1039 (M34). The observations were performed for 22 nights from July 29, 2008 to September 26, 2010. We also made LOAO observations for 10 days from September 18, 2009 to October 30, 2010 to confirm the small variabilties of ${\delta}$ Scuti-type variable stars. In this paper we presented the observational properties of 28 variable stars found in the region. They are seven ${\delta}$ Scuti-type variable stars, two ${\gamma}$ Doradus-type variable stars, four-teen eclipsing binary stars and five semi-long periodic or slow irregular variables, respectively. Only three variables were listed in the GCVS and the rest are newly discovered ones. We have performed multiple-frequency analysis to determine pulsation frequencies of the ${\delta}$ Scuti-type and ${\gamma}$ Doradus-type variable stars, using the discrete Fourier transform and linear least-square fitting methods. We also have derived the periods and amplitudes of 12 eclipsing binaries from the phase fitting method, and presented the light curves of all variable stars.

• Korean Journal of Ecology and Environment
• /
• v.36 no.3 s.104
• /
• pp.344-355
• /
• 2003

Water quality model applicable paddy field was developed using field experiment during 1999 ${\sim}$ 2002. This model involves inputs from fertilization and sediment release as dirac delta function and continuous source function, respectively, and can simulate various processes such as ponded depth, surface drainage, total nitrogen concentration and total phosphorus concentration in a daily basis. The model was calibrated using data collected from field experiments which was irrigated with ground water and validated from field experiments which was irrigated with surface water. The nutrient concentration of surface water depended on the fertilization and dirac delta function can efficiently explain the valiance of nutrient concentration of surface water by fertilizer. As a result of calibration and validation, this model demonstrates good agreement. The model fit efficiencies ($R^2$) of ponded depth, surface concentration of TN and TP were 0.93,0.98 and 0.95, respectively for calibration, and those of TN and TP were 0.99 and 0.70, respectively for validation. We can apply lake and reservoir model to analysis paddy field considered with shallow ponded system, but it will need so many parameters and have much uncertainty. Fortunately, paddy field have a series of cultural practices yearly basis, such as irrigation-fertilization-forced drain-harvest with a similar time , so simple model may explain the mechanism for paddy field. Water quality model for paddy field developed in this study is simply, needs little parameters, but appeared high applicability to evaluate paddy filed drainage. We recommend this model to estimate nutrient loading from paddy field and establish best management practice.

• Asian Pacific Journal of Cancer Prevention
• /
• v.14 no.7
• /
• pp.4255-4259
• /
• 2013

Background: Soft tissue sarcomas (STS) are a heterogeneous group of tumors, and approximately 40-50% of patients with STS develop metastatic disease. The median overall survival of those patients was 12 months and their 5-year survival rate was 8%. Therefore, study on more effective treatment, especially the targeting therapies, is urgently needed. Objective: To evaluate the efficacy and safety of Endostar$^{(R)}$ combined with chemotherapy in patients with advanced STS. Methods: A retrospective case-series study was conducted in Cancer Institute of PLA, Xinqiao Hospital. A total of 71 patients suffering from advanced STS (IIB - IV) were included, of whom 49 cases treated with chemotherapy alone were defined as the control group and the rest 22 cases treated with the traditional chemotherapy combined with Endostar$^{(R)}$ were defined as the test group. The short-term therapeutic effects including objective response rate (ORR), disease control rate (DCR) and safety were evaluated in the two groups. In the follow-up, progression-free survival (PFS) and overall survival (OS) were also observed. Results: In the test and control groups, the ORR was 18.2% and 12.2%, respectively (P=0.767), and the DCR was 86.4% and 61.2%, respectively (P=0.034). The median time to progression in the test and control groups was 120 days and 70 days with significant difference (P = 0.017), while the median overall survival was 452 days and 286 days without significant difference (P=0.503). The one-year survival rate in the test group and control group was 56.2% and 35.4%, respectively, while the two-year survival rate was 30.2% and 26.5%, respectively. No significant difference in the side effects was found between the two groups. Conclusions: Endostar$^{(R)}$ combined with chemotherapy resulted in a higher DCR and longer PFS in the patients with advanced STS, and the toxicity was tolerable.

• Journal of the Korean Geophysical Society
• /
• v.3 no.3
• /
• pp.153-160
• /
• 2000

Crustal velocities around the Korea peninsula are estimated and investigated from eight IGS permanent stations in eastern Asia area. GPS data for the period of May 1995 to December 1991 were analyzed to estimate daily coordinates of each site relative to TAEJ site. The velocity vector of each site is estimated from linear regression analysis with time series of coordinates. As the result, horizontal velocity components for four stations(Tsukuba, Usuda,Taiwan and Shanghai) using thirty-two months data were estimated with the standard error less than 1 mm/year. Our GPS velocity of six sites on the interiors of the Eurasia plate are similar within 1 cm/year with small differences respectively. On the other hand, the velocities of Tsukuba and Usuda showed the great differences from the other six sites. This can be explained by the fact that these two sites are enforced by the surrounding four plates, such as the Pacific, Eurasia, North America and Philipine plate. This study showed that the distance between Korea and Japan is shortened with the rate of 3 cm per year.

• Proceedings of the Korean Society of Crop Science Conference
• /
• 2017.06a
• /
• pp.352-352
• /
• 2017

Recently, the South Korean government encourages the cultivation of upland crops in the paddy field to maintain an adequate level of rice production and then to balance the demand and supply of rice. This is mainly because the rice consumption per capita per year has continued to decline from 135 kg in 1979 to 61.9 kg in 2016, although the rice production was relatively stable. As a result, the rice overproduction became a big social problem. As a part of that, various upland crops such as soybean, maize, minor cereals and forage crops are planted in the paddy field 10 years ago. The cultivation of these crops may settle the problem of short supply and mass import of the crops to some extent. However, a systematic remote observation of upland crops in the paddy field is very scarce. This study investigated the cultivation status of upland crops and any changes of crop harvesting in the paddy field by using an unmanned aerial vehicle (UAV). Also, we analyzed the kind of upland crops and cultivation area in the paddy field by utilizing time series observation images. A fixed wing UAV is used for the investigation. This is because it is easy to use the flight operation and to control flight management software, and it can automatically cope with various emergency states such as a strong wind and battery discharge. The material of UAV is expanded polypropylene, which has an advantage of less equipment damage and risk during takeoff and landing. We acquired observed images in Buljeong-myeon, Goesan-gun, Chungcheongbuk-do, South Korea by using fixed wing UAV in 2015 and 2016. The total investigated area reaches 6,045 ha, and among them the agricultural area was 1,377 ha. For the next step, we created an orthoimage from all images taken using Pix 4D mapper program. According to the results of image analyses in 2015, the paddy field covered total 577 ha (75.9%) with crop plant. The cultivation area of beans, ginseng, maize, tobacco and peach was 256 ha (36.6%), 63 ha (9.2%), 37 ha (5.4%), 31 ha (4.5%) and 27 ha (3.8), respectively. And in 2016, the total covered area was 586 ha (77.1%), and it was comprised of 253 ha (35.5%), 88 ha (12.3%), 29 ha (4.1%), 22 ha (3.1%) and 32 ha (4.5%) in the same order. In this study, we focused on identifying the paddy field which was converted to the cultivation of upland crops by using UAV. And, it has been indicated that the cultivation area of rice decreased from 141 ha in 2015 to 127 ha in 2016, although that of ginseng increased by 25 ha. As a result, it is expected that a lot of paddy field could be replaced by high-income crops such as ginseng and fruit tree (peach) instead of relative low-income rice. More specific and widespread research on the remote sensing in the paddy field needs to be done.

• Journal of the Korean Wood Science and Technology
• /
• v.39 no.1
• /
• pp.75-85
• /
• 2011

The main purpose of this study is to evaluate the potential of producing bioethanol from yellow poplar ($Liriodendron$ $tulipifera$) wood chips by oxalic acid pretreatment and to examine the pretreatment conditions by response surface methodology (RSM). Based on $2^3$ factorial design, adjusted variables were reaction temperature ($^{\circ}C$), residence time (min), and acid loading (g/g), and a series of distinct 15 experimental conditions was organized with duplication at central point (total 16 performances). After pretreatment, simultaneous saccharification and fermentation (SSF) was subjected on solid fraction with yeast strain $Pichia$ $stipitis$. Maximum ethanol yields of the most samples were measured at 72 hours and applied to RSM as a dependent variable. 9.7 g/${\ell}$ of ethanol was produced from the solid pretreated at $180^{\circ}C$ for 40 min with 0.013 g/g of oxalic acid loading. According to the response surface methodology, it was determined that the temperature is the most governing factor via statistic analysis.