• Journal of the Korean Society of Food Science and Nutrition
• /
• v.42 no.9
• /
• pp.1518-1523
• /
• 2013

To estimate the shelf life of red cabbage sprouts (stored at 4 and $10^{\circ}C$), the numbers of total aerobic bacteria were determined during storage. Parameters for the Gompertz model were determined and the shelf life was predicted using a modified Gompertz equation. The estimated shelf lives of red cabbage sprouts packed with polyolefin film and polyamide/polyamide/polyethylene (PA/PA/PE) film at $4^{\circ}C$ were 49.4 and 52.3 h, respectively, whereas those of red cabbage sprouts packed with polyolefin film and PA/PA/PE film at $10^{\circ}C$ were 19.7 and 22.6, respectively. The shelf life prediction equation was appropriate, based on the statistical analysis of the accuracy factor, bias factor, and mean square error. On the other hand, for red cabbage sprouts treated with aqueous $ClO_2$/fumaric acid and UV-C then packed with polyolefin film or PA/PA/PE film, the shelf life was predicted to be longer than 168 h. These results suggest that the combined treatment of aqueous $ClO_2$/fumaric acid and UV-C can be useful for improving microbial safety and extending the shelf life of red cabbage sprouts during storage.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.11 no.2
• /
• pp.204-208
• /
• 2018

In this paper, we proposed and evaluated the time series deep learning prediction model for learning fluctuation pattern of stock price. Recurrent neural networks, which can store previous information in the hidden layer, are suitable for the stock price prediction model, which is time series data. In order to maintain the long - term dependency by solving the gradient vanish problem in the recurrent neural network, we use LSTM with small memory inside the recurrent neural network. Furthermore, we proposed the stock price prediction model using bidirectional LSTM recurrent neural network in which the hidden layer is added in the reverse direction of the data flow for solving the limitation of the tendency of learning only based on the immediately preceding pattern of the recurrent neural network. In this experiment, we used the Tensorflow to learn the proposed stock price prediction model with stock price and trading volume input. In order to evaluate the performance of the stock price prediction, the mean square root error between the real stock price and the predicted stock price was obtained. As a result, the stock price prediction model using bidirectional LSTM recurrent neural network has improved prediction accuracy compared with unidirectional LSTM recurrent neural network.

• Journal of Biosystems Engineering
• /
• v.41 no.4
• /
• pp.357-364
• /
• 2016

Purpose: This study was performed to define the drying characteristics of sorghum by developing thin layer drying equations and evaluating various grain drying equations. Thin layer drying equations lay the foundation characteristics to establish the thick layer drying equations, which can be adopted to determine the design conditions for an agricultural dryer. Methods: The drying rate of sorghum was measured under three levels of drying temperature ($40^{\circ}C$, $50^{\circ}C$, and $60^{\circ}C$) and relative humidity (30%, 40%, and 50%) to analyze the drying process and investigate the drying conditions. The drying experiment was performed until the weight of sorghum became constant. The experimental constants of four thin layer drying models were determined by developing a non-linear regression model along with the drying experiment results. Result: The half response time (moisture ratio = 0.5) of drying, which is an index of the drying rate, was increased as the drying temperature was high and relative humidity was low. When the drying temperature was $40^{\circ}C$ at a relative humidity (RH) of 50%, the maximum half response time of drying was 2.8 h. Contrastingly, the maximum half response time of drying was 1.2 h when the drying temperature was $60^{\circ}C$ at 30% RH. The coefficient of determination for the Lewis model, simplified diffusion model, Page model, and Thompson model was respectively 0.9976, 0.9977, 0.9340, and 0.9783. The Lewis model and the simplified diffusion model satisfied the drying conditions by showing the average coefficient of determination of the experimental constants and predicted values of the model as 0.9976 and Root Mean Square Error (RMSE) of 0.0236. Conclusion: The simplified diffusion model was the most suitable for every drying condition of drying temperature and relative humidity, and the model for the thin layer drying is expected to be useful to develop the thick layer drying model.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.60 no.5
• /
• pp.135-147
• /
• 2018

The objective of this study was to analyze the characteristics of combined 1D/2D inundation simulation of riverside farmland using the Hydrologic Engineering Center - River Analysis System (HEC-RAS). We compared and analyzed inundation simulation results between 1D and combined 1D/2D hydraulic simulation using HEC-RAS. Calibration and validation of stream stage were performed using three rainfall events. The coefficient of determination ($R^2$) and root mean square error (RMSE) between simulated and observed stream stage were 0.935 - 0.957 and 0.250 m - 0.283 m in calibration and validation, respectively. The inundation area showed no significant difference in 1D and combined 1D/2D simulation ($8.48km^2$ in 1D simulation, $8.75km^2$ in combined 1D/2D simulation). The average inundation depth by 1D simulation was 1.4 m deeper than combined 1D/2D simulation. In the lower inundation depth, the inundation area by combined 1D/2D simulation was larger than inundation area by 1D simulation. As the inundation depth increased, the inundation area by 1D simulation became wider. In the case of the 1D/2D combined simulation, low elevation areas along the river bank were inundated widely. Compared to 1D/2D combined simulation, the flood radius in some sections was longer in 1D simulation. In the 1D analysis, because the low altitude riverside farmlands are also assumed to stream, it is calculated that riverside farmlands have the same stage as the mainstream when the stream is overflowed. Therefore, the inundation area seems to be overestimated in those sections. In other regions, the inundation areas tend to be broken depending on overflow by each stream cross-section. In the case of river flooding, the overflow is expected to flow to the lower area depending on the terrain, such as the results of the combined 1D/2D simulation. It is concluded that the results of combined 1D/2D inundation simulation reflected the topographical characteristics of low-lying farmland.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.9 no.3
• /
• pp.278-284
• /
• 2016

Software reliability in the software development process is an important issue. In infinite failure NHPP software reliability models, the fault occurrence rates may have constant, monotonic increasing or monotonic decreasing pattern. In this paper, infinite failures NHPP models that the situation was reflected for the fault occurs in the repair time, were presented about comparing property. Commonly, the software model of the infinite failures using the linear hazard rate distribution software reliability based on intercept parameter was used in business economics and actuarial modeling, was presented for comparison problem. The result is that a relatively large intercept parameter was appeared effectively form. The parameters estimation using maximum likelihood estimation was conducted and model selection was performed using the mean square error and the coefficient of determination. The linear hazard rate distribution model is also efficient in terms of reliability because it (the coefficient of determination is 90% or more) in the field of the conventional model can be used as an alternative model could be confirmed. From this paper, the software developers have to consider intercept parameter of life distribution by prior knowledge of the software to identify failure modes which can be able to help.

• Korean Journal of Remote Sensing
• /
• v.33 no.3
• /
• pp.275-285
• /
• 2017

National Meteorological Satellite Center(NMSC) has produced Sea Surface Temperature (SST) using Communication, Ocean, and Meteorological Satellite(COMS) data since April 2011. In this study, we have developed a new regional COMS SST algorithm optimized within the North-West Pacific Ocean area based on the Multi-Channel SST(MCSST) method and made a composite SST using polar orbit satellites as well as the COMS data. In order to retrieve the optimized SST at Northwest Pacific, we carried out a colocation process of COMS and in-situ buoy data to make coefficients of the MCSST algorithm through the new cloud masking including contaminant pixels and quality control processes of buoy data. And then, we have estimated the composite SST through the optimal interpolation method developed by National Institute of Meteorological Science(NIMS). We used four satellites SST data including COMS, NOAA-18/19(National Oceanic and Atmospheric Administration-18/19), and GCOM-W1(Global Change Observation Mission-Water 1). As a result, the root mean square error ofthe composite SST for the period of July 2012 to June 2013 was $0.95^{\circ}C$ in comparison with in-situ buoy data.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2012.05a
• /
• pp.103-103
• /
• 2012

본 연구에서는 시험포장($1276.6m^2$)에서의 지표피복 BMPs (Best Management Practices) 시나리오를 적용하여 얻은 평균 유출저감율을 HSPF 모델에 적용하여 유역차원에서의 비점원오염 저감효과를 평가하고자 한다. 본 연구에서는 별미천 유역($1.21km^2$)을 대상으로 모형의 적용을 위한 입력자료로 기상자료와 지형자료를 구축하였으며 기상자료로 수원, 양평, 이천 기상관측소 자료를 구축하였으며, 지형자료로 격자크기 2m의 DEM (Digital Elevation Model)과 토지이용도는 2006년 5월 1일 QuickBird 영상을 제공받아 기존 환경부, 건교부, USGS의 토지피복분류체계 및 현장조사를 통하여 QuickBird 영상으로부터 추출 가능한 정밀농업정보에 대한 항목을 결정하였으며, 정사보정된 QuickBird 영상을 스크린 디지타이징 기법(On-Screen Digitizing Method)을 이용하여 총 21개 토지이용항목의 정밀토지이용도를 구축하였다. 실제모니터링으로 측정된 자료를 바탕으로 수위-유량곡선 산정 및 오염부하곡선을 선정, 2011년 6월 8일부터 10월 31일 분석기간으로 HSPF 모델링을 실시하였으며 모의결과 월별 통계에 따른 적용성 분석으로 RMSE (Root Mean Square Error) 는 1.15 ~ 1.76(mm/day), $R^2$는 0.62 ~ 0.78, Nash-Sutcliffe model efficiency (NSE)는 0.62 ~ 0.76로 모의치는 실측치와 유의성이 있는 것으로 분석되었다. 또한, Sediment, T-N, T-P의 $R^2$는 각각 0.72, 0.62, 0.63으로 상관성을 보이는 것으로 분석되었다. 시험포장으로부터 얻어진 event별 볏짚을 이용한 지표피복시나리오적용 후 밭에서의 평균 유출 약 10 % 유출율 감소 조건과 실제 평균 비점원오염 저감효과 89.7 % ~ 99.4 %의 결과로부터 지표피복효과의 침투효과를 HSPF 모델로 적용하기 위해 침투량(INFILT)를 조절하여 평균유출 약 10 %가 감소되는 16.0 mm/hr 값을 선정하였다. 그 결과, Sediment. T-N, T-P의 평균 저감율은 각각 87.2 %, 28.5 %, 85.1 %로 나타났으며 이는 시험포장에서의 실제 평균 비점오염 저감효과 89.7 % ~ 99.4 %에 근접함을 알 수 있었다. 이 결과로부터 침투량 조절에 따른 지표피복(침투짚단)효과는 Sediment, T-P에서 저감효율이 80 % 이상으로 높았지만 T-N은 약 30 %로 낮은 저감율을 보임으로써 저감효과가 크지 않음을 나타냈다.

• Journal of Korea Water Resources Association
• /
• v.51 no.10
• /
• pp.839-852
• /
• 2018

The purpose of this study is to predict monthly agricultural reservoir storage by developing weather data-based Multiple Linear Regression Model (MLRM) with precipitation, maximum temperature, minimum temperature, average temperature, and average wind speed. Using Naïve-Bayes classification, total 1,559 nationwide reservoirs were classified into 30 clusters based on geomorphological specification (effective storage volume, irrigation area, watershed area, latitude, longitude and frequency of drought). For each cluster, the monthly MLRM was derived using 13 years (2002~2014) meteorological data by KMA (Korea Meteorological Administration) and reservoir storage rate data by KRC (Korea Rural Community). The MLRM for reservoir storage rate showed the determination coefficient ($R^2$) of 0.76, Nash-Sutcliffe efficiency (NSE) of 0.73, and root mean square error (RMSE) of 8.33% respectively. The MLRM was evaluated for 2 years (2015~2016) using 3 months weather forecast data of GloSea5 (GS5) by KMA. The Reservoir Drought Index (RDI) that was represented by present and normal year reservoir storage rate showed that the ROC (Receiver Operating Characteristics) average hit rate was 0.80 using observed data and 0.73 using GS5 data in the MLRM. Using the results of this study, future reservoir storage rates can be predicted and used as decision-making data on stable future agricultural water supply.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.36 no.9
• /
• pp.1065-1071
• /
• 2012

Wind energy is becoming one of the most preferable alternatives to conventional sources of electric power that rely on fossil fuels. For stable electric power generation, constant rotating speed control of a wind turbine is performed through pitch control and stall control of the turbine blades. Recently, variable pitch control has been implemented in modern wind turbines to harvest more energy at variable wind speeds that are even lower than the rated one. Although wind turbine pitch controllers are currently optimized using a step response via the Ziegler-Nichols auto-tuning process, this approach does not satisfy the requirements of variable pitch control. In this study, the variable pitch controller was optimized by a genetic algorithm using a neural network model that was constructed by the Latin Hypercube sampling method to improve the Ziegler-Nichols auto-tuning process. The optimized solution shows that the root mean square error, rise time, and settle time are respectively improved by more than 7.64%, 15.8%, and 15.3% compared with the corresponding initial solutions obtained by the Ziegler-Nichols auto-tuning process.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.24 no.1
• /
• pp.15-19
• /
• 2020

Purpose Generative Adversarial Network(GAN) is one of deep learning technologies. This is a way to create a real fake image after learning the real image. In this study, after acquiring artificial intelligence images through GAN, We were compared and evaluated with real scan time images. We want to see if these technologies are potentially useful. Materials and Methods 30 patients who underwent ¹⁸F-FDG Brain PET/CT scanning at Severance Hospital, were acquired in 15-minute List mode and reconstructed into 1,2,3,4,5 and 15minute images, respectively. 25 out of 30 patients were used as learning images for learning of GAN and 5 patients used as verification images for confirming the learning model. The program was implemented using the Python and Tensorflow frameworks. After learning using the Pix2Pix model of GAN technology, this learning model generated artificial intelligence images. The artificial intelligence image generated in this way were evaluated as Mean Square Error(MSE), Peak Signal to Noise Ratio(PSNR), and Structural Similarity Index(SSIM) with real scan time image. Results The trained model was evaluated with the verification image. As a result, The 15-minute image created by the 5-minute image rather than 1-minute after the start of the scan showed a smaller MSE, and the PSNR and SSIM increased. Conclusion Through this study, it was confirmed that AI imaging technology is applicable. In the future, if these artificial intelligence imaging technologies are applied to nuclear medicine imaging, it will be possible to acquire images even with a short scan time, which can be expected to reduce artifacts caused by patient movement and increase the efficiency of the scanning room.