• International Journal of Contents
• /
• 제4권2호
• /
• pp.7-12
• /
• 2008

In this paper, we present a real-time physical simulation model of water surfaces with a novel method to represent the water mass flow in full three dimensions. In a physical simulation model, the state of the water surfaces is represented by a set of physical values, including height, velocity, and the gradient. The evolution of the velocity field in previous works is handled by a velocity solver based on the Navier-Stokes equations, which occurs as a result of the unevenness of the velocity propagation. In this paper, we integrate the principle of the mass conservation in a fluid of equilateral density to upgrade the height field from the unevenness, which in mathematical terms can be represented by the divergence operator. Thus the model generates waves induced by horizontal velocity, offering a simulation that puts forces added in all direction into account when calculating the values for height and velocity for the next frame. Other effects such as reflection off the boundaries, and interactions with floating objects are involved in our method. The implementation of our method demonstrates to run with fast speed scalable to real-time rates even for large simulation domains. Therefore, our model is appropriate for a real-time and large scale water surface simulation into which the animator wishes to visualize the global fluid flow as a main emphasis.

• 한국시뮬레이션학회논문지
• /
• 제13권2호
• /
• pp.13-21
• /
• 2004

The thermal-hydraulic model ARTS which was based on the RETRAN-3D code adopted in the domestic full-scope power plant simulator which was provided in 1998 by KEPRI. Since ARTS is a generalized code to model the components with control volumes, the smaller time-step size should be used even if converged solution could not get in a single volume. Therefore, dedicated models which do not force to reduce the time-step size are sometimes more suitable in terms of a real-time calculation and robustness. In the case of PRT(Pressurizer Relief Tank) model, it is consist of subcooled water in bottom and non-condensable gas in top. The sparger merged under subcooled water enhances condensation. The complicated thermal-hydraulic phenomena such as condensation, phase separation with existence of non-condensable gas makes difficult to simulate. Therefore, the PRT volume can limit the time-step size if we model it with a general control volume. To prevent the time-step size reduction due to convergence failure for simulating this component, we developed a dedicated model for PRT. The dedicated model was expected to provide substantially more accurate predictions in the analysis of the system transients. The results were resonable in terms of accuracy, real-time simulation, robustness and education of operators, complying with the ANSI/ANS-3.5-1998 simulator software performance criteria and RETRAN-3D results.

• 정보처리학회논문지:컴퓨터 및 통신 시스템
• /
• 제12권2호
• /
• pp.53-60
• /
• 2023

최근 실시간 스트리밍 플랫폼을 기반으로 한 다양한 멀티미디어 컨텐츠의 수요량과 트래픽 양이 급격히 증가하고 있는 추세이다. 본 논문에서는 실시간 스트리밍 서비스의 품질을 향상시키기 위해서 실시간 스트리밍 트래픽을 예측한다. 네트워크 트래픽을 예측하기 위해 통계적 모형을 활용하였으나, 실시간 스트리밍 트래픽은 매우 동적으로 변화함에 따라 통계적 모형보다는 순환 신경망 기반 딥러닝 모델이 적합하다. 따라서, 실시간 스트리밍 트래픽을 수집, 정제 후 Vanilla RNN, LSTM, GRU, Bi-LSTM, Bi-GRU 모델을 활용하여 예측하며, 각 모델의 학습 시간, 정확도를 측정하여 비교한다.

• 한국도로학회논문집
• /
• 제10권1호
• /
• pp.155-164
• /
• 2008

최근 들어 널리 확대되는 ITS시설과 장비의 효율적인 활용을 위해서는 실시간 통행시간예측과 같은 핵심기술의 개발이 매우 중요하다. 실시간 통행시간 예측기술을 통해 도로 이용자에게 통행시간 정보를 제공해줌으로써 운전자가 정보에 기반을 둔 선택을 할 수 있게 되며, 이에 따라 도로이용자의 통행 효용 및 도로시설 이용효율을 극대화할 수 있다. 본 연구에서는 자료 성질이 다른 VDS정보와 AVI정보를 하나의 틀 안에서 융합하는 모형을 제시하였으며, 이를 통해 VDS정보의 실시간성과 AVI 정보의 현실반영 특성을 반영한 통행시간추정이 가능하게 되었다. 또한 실시간 교통정보 제공을 위해 알고리즘의 수치연산 처리량이 실시간 현장 적용이 가능한 수준이 되도록 모형을 개발하였다. 본 연구에서 제시된 통행시간 추정 모형의 적용가능성과 신뢰성은 실제 국도 교통량자료를 이용하여 검증하였다. 실험결과에 따르면 본 모형은 연산처리 효율성이 매우 우수하여 실시간 운영 이 가능할 뿐만 아니라 실시간으로 수집되는 검지정 보를 사용함으로써 예측의 정확도를 향상시켰다. 특히 본 연구에서 제시한 이질적 자료의 융합방법은 향후 새로운 형태의 검지자료를 활용하는데 있어서 매우 유용하게 사용될 수 있을 것으로 기대된다.

• Environmental Sciences Bulletin of The Korean Environmental Sciences Society
• /
• 제1권1호
• /
• pp.35-43
• /
• 1997

Handling the emergency problems such as Chemobyl accident require real time prediction of pollutants dispersion. One-point real time sounding at pollutant source and simple model including turbulent-radiation process are very important to predict dispersion at real time. The stability categories obtained by one-dimensional numerical model (including PBL dynamics and radiative process) are good agreement with observational data (Golder, 1972). Therefore, the meteorological parameters (thermal, moisture and momentum fluxes; sensible and latent heat; Monin-Obukhov length and bulk Richardson number; vertical diffusion coefficient and TKE; mixing height) calculated by this model will be useful to understand the structure of stable boundary layer and to handling the emergency problems such as dangerous gasses accident. Especially, this simple model has strong merit for practical dispersion models which require turbulence process but does not takes long time to real predictions. According to the results of this model, the urban area has stronger vertical dispersion and weaker horizontal dispersion than rural area during daytime in summer season. The maximum stability class of urban area and rural area are 'A' and 'B' at 14 LST, respectively. After 20 LST, both urban and rural area have weak vertical dispersion, but they have strong horizontal dispersion. Generally, the urban area have larger radius of horizontal dispersion than rural area. Considering the resolution and time consuming problems of three dimensional grid model, one-dimensional model with one-point real sounding have strong merit for practical dispersion model.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제51권1호
• /
• pp.39-42
• /
• 2002

For the purpose of modeling EEG signal which has nonstationary and nonlinear dynamic characteristics, this paper propose a state feedback real time recurrent neural network model. The state feedback real time recurrent neural network is structured to have memory structure in the state of hidden layers so that it has arbitrary dynamics and ability to deal with time-varying input through its own temporal operation. For the model test, Mackey-Glass time series is used as a nonlinear dynamic system and the model is applied to the prediction of three types of EEG, alpha wave, beta wave and epileptic EEG. Experimental results show that the performance of the proposed model is better than that of other neural network models which are compared in this paper in some view points of the converging speed in learning stage and normalized mean square error for the test data set.

• 전력전자학회:학술대회논문집
• /
• 전력전자학회 2016년도 전력전자학술대회 논문집
• /
• pp.357-358
• /
• 2016

This paper presents a dynamic, electric-thermal model for a photovoltaic (PV) cell that combines electrical and thermal parameters. In this model, the irradiance and ambient temperature are used to calculate the PV cell temperature based on a five-layer thermal model. The cell temperature is then used in the electrical model to accurately adjust the PV cell output electrical characteristics and power. A custom experimental setup was built to test and verify the electrical and thermal characteristics of the PV cell and its surrounding layers. The electric-thermal model is validated using experimental data in realistic scenarios. This PV model can be scaled up and used to simulate PV systems in wide variety of applications, extreme environmental conditions, and fault conditions in real-time.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1988년도 한국자동제어학술회의논문집(국내학술편); 한국전력공사연수원, 서울; 21-22 Oct. 1988
• /
• pp.466-470
• /
• 1988

A tunable reduced-order distillation model is proposed for real-time applications. To develop the model, a binary distillation column with MaCabe-Thiele assumptions was considered first and then the governing equations for the column were reduced to a simplified vector differential equations using the collocation method combined with cubic spline interpolation function. The final reduced-order model has four tuning parameters, relative volatilities and liquid holdups for rectifying and stripping sections, respectively. To assess the applicability of the developed model,the real-time adjustment of the model was tried by recursively updating the tuning parameters using the BKF algorithm. As a result, it was found that the reduced-model follows the simulated distillation process very closely as the parameters are improved.

• 한국정보과학회논문지:소프트웨어및응용
• /
• 제31권6호
• /
• pp.773-783
• /
• 2004

본 연구는 스테이트차트의 실시간 검증을 위한 알고리즘을 제안한다. 스테이트차트는 실시간 반응형 시스템의 명세에 많이 사용되고 있으며 동기적과 비동기적인 두개의 시간모델을 지원한다. 그러나 기존의 스테이트차트에 대한 실시간 검증 방법은 비동기적 시간 모델과 호환되지 않거나, 변수를 모델에 추가함으로써 모텔의 상태 공간을 증가시키는 단점이 있었다. 우리는 기존의 모델 체킹 알고리즘을 확장하여 이러한 문제점을 해결하였다. 확장된 알고리즘은 시간을 증가시키는 전이만을 고려하기 때문에 스테이트차트의 두 가지 시간 모델에 모두 사용할 수 있으며, 시간의 계산이 알고리즘 내부적으로 이루어지기 때문에 모델에 변수를 추가할 필요가 없어 상태공간을 증가시키지 않는다. 본 연구는 이러한 알고리즘을 토대로 기존의 모델 체커인 NuSMV를 확장하였으며, 사례 연구를 통하여 그 유용성을 확인하였다.

• International Journal of Advanced Culture Technology
• /
• 제11권1호
• /
• pp.325-331
• /
• 2023

In this paper, we propose a mixed neural network structure of CNN and LSTM that can be used to detect or predict odor occurrence, which is most required in manufacturing industry or real life, using odor complex sensors. In addition, the proposed learning model uses a complex odor sensor to receive four types of data such as hydrogen sulfide, ammonia, benzene, and toluene in real time, and applies this data to an inference model to detect and predict odor conditions. The proposed model evaluated the prediction accuracy of the learning model through performance indicators according to accuracy, and the evaluation result showed an average performance of 94% or more.