• Journal of the Korean earth science society
• /
• v.32 no.2
• /
• pp.190-199
• /
• 2011

A photovoltaic energy map that included a topography effect on the Korean peninsula was developed using the Gangneung-Wonju National University (GWNU) solar radiation model. The satellites data (MODIS, OMI and MTSAT-1R) and output data from the Regional Data Assimilation Prediction System (RDAPS) model by the Korea Meteorological Administration (KMA) were used as input data for the GWNU model. Photovoltaic energy distributions were calculated by applying high resolution Digital Elevation Model (DEM) to the topography effect. The distributions of monthly accumulated solar energy indicated that differences caused by the topography effect are more important in winter than in summer because of the dependency on the solar altitude angle. The topography effect on photovoltaic energy is two times larger with 1 km resolution than with 4 km resolution. Therefore, an accurate calculation of the solar energy on the surface requires high-resolution topological data as well as high quality input data.

• The KSFM Journal of Fluid Machinery
• /
• v.12 no.3
• /
• pp.44-52
• /
• 2009

Small scale steam turbines are used as mechanical drivers in chemical process plant or power generators. In this study, a design technology was developed for a 100kW class steam turbine which will be used for removing $CO_2$ from the emission gas on a reheated cycle system. This turbine is operated at a low inlet total pressure of $5\;kgf/cm^2$. It consists of two stages and operates at the partial admission. For the meanline analysis, a performance prediction method was developed and it was validated through the performances on the operating small steam turbines which are using at plants. Their results showed that the output power was predicted within 10% deviation although the steam turbines adopted in this analysis were operated at different flow conditions and rotor size. The turbine blades was initially designed based on the computed results obtained from the meanline analysis. A supersonic nozzle was designed on the basis of the operating conditions of the turbine, and the first stage rotor was designed using a supersonic blade design method. The stator and second stage rotor was designed using design parameters for the blade profile. Finally, Those blades were iteratively modified from the flow structures obtained from the three-dimensional flow analysis to increase the turbine performance. The turbine rotor system was designed so that it could stably operate by 76% separation margin with tilting pad bearings.

• Phonetics and Speech Sciences
• /
• v.1 no.4
• /
• pp.203-207
• /
• 2009

Electrically evoked compound action potentials (ECAP) have originated from the distal end of the auditory nerve. ECAP are characterized as the difference between the clearly large trough (N) and the following positive peak (P). N-wave occurs around $200-400\;{\mu}s$ after stimulus onset and P-wave at around $400-800\;{\mu}s$. Contrary to expectations, positive peaked ECAP (pp-ECAP) was dominated by a relatively large-amplitude positive following negative peak. pp-ECAP can be recorded from the sites on or near the surgically exposed nerve trunk in animal models and/or in cases of monophasic stimulation. This study will provide the causes of the appearance of pp-ECAP in cases of cochlear implant recipients using imaging studies and medical records and statistically analysis between N-P and P-N on the amplitude input-output function (amp-I/O) for the prediction of the possibilities of clinical tools. Thirteen children participated in the study and received a Cochlear CI-24RE (CA). ECAP was recorded using auto-NRT (Cochlear Ltd., Australia) at four to five weeks post surgery. pp-ECAP was measured from 36 electrodes and typical ECAP from 220 electrodes. There was no abnormality in the imaging study and operation finding in patients with typical ECAP. pp-ECAP was found at the inner ear anormaly and ossification in imaging study and gel-state inner ear fluid was observed in the operation finding. The amplitude of pp-ECAP increased depending on current intensities, but amp-I/O increase more gradually than in the case of typical ECAP (p=0.003). pp-ECAP is antidromic potential which can record from the inner ear anormaly and ossified cochlear. Amp-I/O also depends on current intensity as well typical ECAP. These results provide a useful tool for audiological evaluation for the spiral ganglion cell status to the value of pp-ECAP.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.17 no.2 s.105
• /
• pp.101-109
• /
• 2006

A broadband, DC to 40 GHz 5-bit MMIC digital attenuator has been developed. The ultra broadband attenuator has been achieved by adding transmission lines in the conventional Switched-T attenuator and optimizing the transmission line parameters. Momentum simulation was performed in design for accurate performance prediction at high frequencies and Monte Carlo analysis was applied to verify the performance stability against the MMIC process variation. The attenuator has been fabricated with $0.15\;{\mu}m$ GaAs pHEMT process. This attenuator has 1 dB resolution and 23 dB dynamic ranges. High attenuation accuracy has been achieved over all attenuation ranges and 40 GHz bandwidth with the reference state insertion loss of less than 6 dB at 20 GHz. The input and output return losses of the attenuator are better than 14 dB over all attenuation states and frequencies. The measured IIP3 of the attenuator is 33 dBm.

• Journal of Sensor Science and Technology
• /
• v.9 no.2
• /
• pp.127-133
• /
• 2000

We propose a novel monitoring system for a recycling piggery slurry treatment system through neural networks. Here we tried to model treatment process for each tank(influent, fermentation, aeration, first sedimentation and fourth sedimentation tanks) in the system based on population densities of heterotrophic and lactic acid bacteria. Principle component analysis(PCA) was first applied to identify a relation between input(microbial densities and parameters for the treatment) and output, and then multilayer neural networks were employed to model the treatment process for each tank. PCA filtration of input data as microbial densities was found to facilitate the modeling procedure for the system monitoring even with a relatively lower number of input. Neural networks independently trained for each treatment tank and their subsequent combinatorial data analysis allowed a successful prediction of the treatment system for at least two days.

• Journal of Broadcast Engineering
• /
• v.12 no.4
• /
• pp.303-310
• /
• 2007

Motion Estimation (ME) is a core part of most Video compression systems since it affects directly the output video quality and the encoding time. The most basic method of ME, Full Search (FS) gives the highest visual quality but also has the problem of significant computational load. To solve this problem, many fast algorithm has been proposed. Among them, MVFAST and PMVFAST show impressive results in video quality and the computational load by using the correlation between motion vectors of adjacent blocks. In particular, PMVFAST reduces search points dramatically and also gives very high video quality by using the median predictor. In this paper, we propose a new algorithm that uses the redefined median predictor which reduces the number of search points and yields a high visual quality by reducing the number of thresholds and early termination conditions.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2006.05a
• /
• pp.149-153
• /
• 2006

Precipitation is the most important component and critical to the study of water and energy cycle. This study investigates the propagation of precipitation retrieval uncertainty in the simulation of hydrologic variables for varying spatial resolution on two different vegetation cover. We explore two remotely sensed rain retrievals (space-borne IR-only and radar rainfall) and three spatial grid resolutions. An offline Community Land Model (CLM) was forced with in situ meteorological data In turn, radar rainfall is replaced by the satellite rain estimates at coarser resolution $(0.25^{\circ},\;0.5^{\circ}\;and\;1^{\circ})$ to determine their probable impact on model predictions. Results show how uncertainty of precipitation measurement affects the spatial variability of model output in various modelling scales. The study provides some intuition on the uncertainty of hydrologic prediction via interaction between the land surface and near atmosphere fluxes in the modelling approach.

• Journal of the Korean Institute of Gas
• /
• v.1 no.1
• /
• pp.41-48
• /
• 1997

The equation of state is widely utilized as a simple model for the prediction of gas properties. There are several equations of state and they often make diverse and hard to believe output of gas properties. In this study, We show a reliability of equation of state for nitrogen, oxygen and argon in pressure range from 1 bar to 30 bar and temperature range from liquefaction to room temperature. We use three equations of state such as Soave-Redlich-Kwong, Peng-Robinson and BWR-LS' equation of state which provided in the Aspen plus. The results were compared with literatures and virial equation. Finally, We report the differences of process calculation of distillation column and expansion turbine in cryogenic air separation plant with change of equation of state.

• Journal of KIISE
• /
• v.44 no.6
• /
• pp.607-612
• /
• 2017

This paper presents an attempt to apply Deep Learning technology to solve the problem of forecasting floods in urban areas. We employ Recurrent Neural Networks (RNNs), which are suitable for analyzing time series data, to learn observed data of river water and to predict the water level. To test the model, we use water observation data of a station in the Trinity river, Texas, the U.S., with data from 2013 to 2015 for training and data in 2016 for testing. Input of the neural networks is a 16-record-length sequence of 15-minute-interval time-series data, and output is the predicted value of the water level at the next 30 minutes and 60 minutes. In the experiment, we compare three Deep Learning models including standard RNN, RNN trained with Back Propagation Through Time (RNN-BPTT), and Long Short-Term Memory (LSTM). The prediction quality of LSTM can obtain Nash Efficiency exceeding 0.98, while the standard RNN and RNN-BPTT also provide very high accuracy.

• The Journal of the Korea Contents Association
• /
• v.10 no.6
• /
• pp.166-174
• /
• 2010

Based on the theoretical results that multi-layer feedforward neural networks with enough hidden nodes are universal approximators, we usually use three-layer MLP's(multi-layer perceptrons) consisted of input, hidden, and output layers for many application problems. However, this conventional three-layer architecture of MLP shows poor generalization performance in some applications, which are complex with various features in an input vector. For the performance improvement, this paper proposes a hierarchical architecture of MLP especially when each part of inputs has a special information. That is, one input vector is divided into sub-vectors and each sub-vector is presented to a separate MLP. These lower-level MLPs are connected to a higher-level MLP, which has a role to do a final decision. The proposed method is verified through the simulation of protein disorder prediction problem.