• Journal of the Korean Institute of Intelligent Systems
• /
• v.17 no.4
• /
• pp.563-568
• /
• 2007

This paper presents a methodology for predicting nonlinear time series based on the neural network with weighted fuzzy membership functions (NEWFM). The degree of classification intensity is obtained by bounded sum of weighted fuzzy membership functions extracted by NEWFM, then weighted average defuzzification is used for predicting nonlinear time series. The experimental results demonstrate that NEWFM has the classification capability of 92.22% against the target class of GDP. The time series created by NEWFM model has a relatively close approximation to the GDP which is a typical business cycle indicator, and has been proved to be a useful indicator which has the turning point forecasting capability of average 12 months in the peak point and average 6 months in the trough point during 5th to 8th cyclical period. In addition, NEWFM measures the efficiency of the economic indexes by the feature selection and enables the users to forecast with reduced numbers of 7 among 10 leading indexes while improving the classification rate from 90% to 92.22%.

• Journal of KIISE:Computer Systems and Theory
• /
• v.32 no.7
• /
• pp.377-392
• /
• 2005

A lot of studies to measure and analyze the system performance have been done in areas such as system modeling, performance measurement, monitoring, and performance prediction since the advent of a computer system. Studies on a framework to unify the performance related areas have rarely been performed although many studies in the various areas have been done, however. In the case of TMO(Time-Triggered Message-Triggered Object), a real-time programming model, it hardly provides tools and frameworks on the performance except a simple run-time monitor. So it is difficult to analyze the performance of the real-time system and the process based on TMO. Thus, in this paper, we propose a framework for the dynamic analysis of the real-time system based on TMO, TDAF(TMO based Dynamic Analysis Framework). TDAF treats all the processes for the performance measurement and analysis, and Provides developers with more reliable information systematically combining a load model, a performance model, and a reporting model. To support this framework, we propose a load model which is extended by applying TMO model to the conventional one, and we provide the load calculation algorithm to compute the load of TMO objects. Additionally, based on TMO model, we propose performance algorithms which implement the conceptual performance metrics, and we present the reporting model and algorithms which can derive the period and deadline for the real-time processes based on the load and performance value. In last, we perform some experiments to validate the reliability of the load calculation algorithm, and provide the experimental result.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.22 no.2
• /
• pp.79-91
• /
• 2020

Accurate assessment of greenhouse gas emissions is a cornerstone of every climate change response study, and reliable assessment of greenhouse gas emission data is being used as a practical basis for the entire climate change prediction and modeling studies. Essential, fundamental technologies for estimating greenhouse gas emissions include an on-site monitoring technology, an evaluation methodology of uncertainty in emission factors, and a verification technology for reductions. The closed chamber method is being commonly used to measure gas fluxes between soil-vegetation and atmosphere. This method has the advantages of being simple, easily available and economical. This study presented the technical bases of the closed chamber method for measuring methane fluxes from a rice paddy. The methane fluxes from rice paddies occupy the largest portion of a single source of greenhouse gas in the agricultural field. We reviewed the international and the domestic studies on automated chamber monitoring systems that have been developed from manually operated chambers. Based on this review, we discussed scientific concerns on chamber methods with a particular focus on quality control for improving measurement reliability of field data.

• Tunnel and Underground Space
• /
• v.30 no.1
• /
• pp.39-62
• /
• 2020

In this study, Barcelona Basic Model (BBM) was implemented into TOUGH2-MP/FLAC3D for the numerical analysis of coupled thermo-hydro-mechanical (THM) behavior of unsaturated soils and the prediction of long-term behaviors. Similar to the methodology described in a previous study for the implementation of BBM into TOUGH-FLAC, the User Defined Model (UDM) of FLAC based on the Modified Cam Clay Model (MCCM) and the FISH function of FLAC3D were used to extend the existing MCCM module in FLAC3D for the implementation of BBM into TOUGH2-MP/FLAC3D. In the developed BBM module in TOUGH2-MP/FLAC3D, the plastic strains due to change in suction increase (SI) in addition to mean effective stress are calculated. In addition to loading-collapse (LC) yield surface, suction increase (SI) yield surface is changed by hardening rules in the developed BBM module. Several numerical simulations were conducted to verify and validate the implementation of BBM: using an example presented in the FLAC3D manual for the standard MCCM, simulation results using COMSOL, and experimental data presented in SKB Reports. In addition, the developed BBM analysis module was validated by simultaneously performing a series of modeling tests that were performed for the validation of the Quick tools developed for the purpose of effectively deriving BBM parameters, and by comparing the Quick tools and Code_Bright results reported in a previous study.

• Journal of Korea Water Resources Association
• /
• v.41 no.10
• /
• pp.995-1007
• /
• 2008

Studies on modeling the rainfall-runoff relationship which shows nonlinear trend strongly use artificial neural networks theory not only for the prediction but also for the characteristics analysis of the data used by pattern classification. For the pattern classification, the results from Self-Organizing Map (SOM) mention that the map size and array for the SOM training have significantly influenced on the SOM performance. Since there is no deterministic method or theoretical equation to determine the number of rows and columns for the map size, hexagonal array is generally used for the map array. Therefore, this study present a determination of the optimized map structure for the rainfall-runoff analysis in Naju station considering the map size and array simultaneously which can represent the classified characterization of rainfall-runoff relationship. The result showed that the map size of 20$\times$16 hexagonal array with 8-clustered patterns was selected as an appropriate map structure for rainfall-runoff analysis in Naju station.

• Journal of the Korea Society for Simulation
• /
• v.25 no.3
• /
• pp.107-116
• /
• 2016

The recent diversification in terms of the scope and techniques used for simulations has highlighted the importance of analyzing state of the art trends and applying these for educational and study purposes. While qualitative methods such as literature research or experts' assessments have previously been used, such methods are in fact likely to reflect the subjective viewpoint of experts, and to involve too much time and money for the results obtained. For the purpose of an objective analysis, a quantitative analysis that included the examination of topics found in domestic academic journal articles was conducted in the present study. In this regard, simulation was found to be most actively used domestically in the electrical and electronic fields. In addition, simulation was also found to be employed for the purpose of education and entertainment in the social sciences. The results of this study are expected to help to facilitate the prediction of the direction of the development of not only the Korea Society for Simulation, but also domestic simulation studies. This study also raises the possibility of applying text mining to trend analysis, and proves that it can be a useful method for deriving future key topics and helping experts' decisions regarding quantitative data.

• The Journal of Bigdata
• /
• v.6 no.1
• /
• pp.51-62
• /
• 2021

In the field of medical artificial intelligence, there have been a lot of researches on disease prediction and classification algorithms that can help doctors judge, but relatively less interested in artificial intelligence that can help medical consumers acquire and judge information. The fact that more than 150,000 questions have been asked about which hospital to go over the past year in NAVER portal will be a testament to the need to provide medical information suitable for medical consumers. Therefore, in this study, we wanted to establish a classification model that classifies 8 medical subjects for symptom text directly described by patients which was collected from NAVER portal to help consumers choose appropriate medical subjects for their symptoms. In order to ensure the validity of the data involving patients' subject matter, we conducted similarity measurements between objective symptom text (typical symptoms by medical subjects organized by the Seoul Emergency Medical Information Center) and subjective symptoms (NAVER data). Similarity measurements demonstrated that if the two texts were symptoms of the same medical subject, they had relatively higher similarity than symptomatic texts from different medical subjects. Following the above procedure, the classification model was constructed using a ridge regression model for subjective symptom text that obtained validity, resulting in an accuracy of 0.73.

• Ecology and Resilient Infrastructure
• /
• v.8 no.4
• /
• pp.165-178
• /
• 2021

We studied driftwood behaviors including generation and deposition in a tsunami using a numerical simulation. We used an integrated two-dimensional numerical model, which included a driftwood dynamics model. The study area was Sendai, Japan. Observation data collected by Inagaki et al. (2012) were used to verify the simulation results by comparing them with driftwood deposition patterns. A simplified model was developed to consider the threshold of driftwood generation by the drag force of water flows. To consider the volume of driftwood generated, we estimated the total wood number in the study area using Google Earth. Therefore, we simulated more than 13,000 pieces of driftwood that were generated and transported inland from approximately 300,000 trees that were growing in the forest. The final distribution of the driftwood was similar to the observation data. The reproducibility of the generation and deposition patterns of driftwood showed good agreement in terms of longitudinal deposition pattern. In the future, a sensitivity analysis on driftwood parameters, such as the size of the wood, boundary conditions, and grid size, will be implemented to predict the travel patterns of driftwood. Such modeling will be a useful methodology for disaster prediction based on water flow and driftwood.

• Korean Chemical Engineering Research
• /
• v.57 no.2
• /
• pp.274-282
• /
• 2019

In recent years, big data analysis has been expanded to include automatic control through reinforcement learning as well as prediction through modeling. Research on the utilization of image data is actively carried out in various industrial fields such as chemical, manufacturing, agriculture, and bio-industry. In this paper, we applied NASNet, which is an AutoML reinforced learning algorithm, to DeepU-Net neural network that modified U-Net to improve image semantic segmentation performance. We used BRATS2015 MRI data for performance verification. Simulation results show that DeepU-Net has more performance than the U-Net neural network. In order to improve the image segmentation performance, remove dropouts that are typically applied to neural networks, when the number of kernels and filters obtained through reinforcement learning in DeepU-Net was selected as a hyperparameter of neural network. The results show that the training accuracy is 0.5% and the verification accuracy is 0.3% better than DeepU-Net. The results of this study can be applied to various fields such as MRI brain imaging diagnosis, thermal imaging camera abnormality diagnosis, Nondestructive inspection diagnosis, chemical leakage monitoring, and monitoring forest fire through CCTV.

• Atmosphere
• /
• v.29 no.1
• /
• pp.87-103
• /
• 2019

This study introduces the overall characteristics of LOVECLIM version 1.3, the earth system model of intermediate complexity (EMIC), including the installation and operation processes by conducting two kinds of past climate simulation. First climate simulation is the equilibrium experiment during the mid-Holocene (6,000 BP), when orbital parameters were different compared to those at present. The overall accuracy of simulated global atmospheric fields by LOVECLIM is relatively lower than that in Coupled Model Intercomparison Project phase 5 (CMIP5) and Paleoclimate modelling Intercomparison Project phase 3 (PMIP3) simulations. However, surface temperature over the globe, the 800 hPa meridional wind over the mid-latitude coastal region, and the 200 hPa zonal wind from LOVECLIM show similar spatial distribution to those multi-model mean of CMIP5/PMIP3 climate models. Second one is the transient climate experiment from mid-Holocene to present. LOVECLIM well captures the major differences in surface temperature between preindustrial and mid-Holocene simulations by CMIP5/PMIP3 multi-model mean, even though it was performed with short integration time (i.e., about four days in a single CPU environment). In this way, although the earth system model of intermediate complexity has a limit due to its relatively low accuracy, it can be a very useful tool in the specific research area such as paleoclimate.