• Journal of Institute of Control, Robotics and Systems
• /
• v.10 no.12
• /
• pp.1295-1304
• /
• 2004

In this paper, the Bayesian recurrent neural network is proposed to predict time series data. A neural network predictor requests proper learning strategy to adjust the network weights, and one needs to prepare for non-linear and non-stationary evolution of network weights. The Bayesian neural network in this paper estimates not the single set of weights but the probability distributions of weights. In other words, the weights vector is set as a state vector of state space method, and its probability distributions are estimated in accordance with the particle filtering process. This approach makes it possible to obtain more exact estimation of the weights. In the aspect of network architecture, it is known that the recurrent feedback structure is superior to the feedforward structure for the problem of time series prediction. Therefore, the recurrent neural network with Bayesian inference, what we call Bayesian recurrent neural network (BRNN), is expected to show higher performance than the normal neural network. To verify the proposed method, the time series data are numerically generated and various kinds of neural network predictor are applied on it in order to be compared. As a result, feedback structure and Bayesian learning are better than feedforward structure and backpropagation learning, respectively. Consequently, it is verified that the Bayesian reccurent neural network shows better a prediction result than the common Bayesian neural network.

• The Korean Journal of Applied Statistics
• /
• v.22 no.5
• /
• pp.961-969
• /
• 2009

Water absorption in water-cooled generator stator windings can cause serious accidents such as insulation breakdown and it brings a generator to the unexpected sudden outage. Accordingly, it is important to diagnose the water absorption of them in the effective operation of power plant. Especially, the capacitance value which is measured for diagnosis is very small so the special diagnosis methods like stochastic theory are needed. KEPRI developed the water absorption test equipment and diagnosis technology for them. In this paper we propose that water absorption test of generator stator windings using probability distributions. The proposed diagnosis technology is applied to the real system and the results of water absorption test for stator windings are agreed to them of water leak test.

• Current Research on Agriculture and Life Sciences
• /
• v.15
• /
• pp.53-67
• /
• 1997

The objective of this study is to provide with the hydro-meteological and probabilistic characteristics of the storms of typhoons that have been passed through the Korean peninsula during the last twenty-three years since 1961. The paths and intensities of the typhoons were analyzed. Fifty weather stations were selected and the rainfall data during typhoon periods were collected. Rainfall data were analyzed for the patterns and probabilistic distributions. The results were presented to describe the areal distributions of probabilistic characteristics. The results obtained from this study can be summarized as follows: 1. The most frequent typhoon path that has passed through the Korean peninsula was type E, followed by types CWE, W, WE, and S. The most frequent typhoon intensity was type B, followed by A, super A, and e types, respectively. 2. The third quartile typhoon rainfall patterns appear most frequently followed by the second, first, and last quartiles, respectively, in Seoul, Pusan, Taegu, Kwangju and Taejon. The single typhoon rainfalls with long rainfall durations tended to show delayed type rainfall patterns predominantly compared to the single rainfalls with short rainfall durations. 3. The most frequent probabilistic distribution of typhoon rainfall event is Pearson type-III, followed by Two-parameter lognormal distribution, and Type-I extremal distribution. 4. The most frequent probability distribution model of seashore location was Pearson type-III distribution. The most frequent probability distribution model of inland location was two parameter lognormal distribution.

• Journal of Environmental Science International
• /
• v.10 no.3
• /
• pp.195-200
• /
• 2001

The personal exposures of nitrogen dioxide(NO$_2$), microenvironmental levels and daily time activity patterns on Seoul subway station workers were measured from February 10 to March 12, 1999. Personal NO$_2$exposure for 24 hours were 29.40$\pm$9.75 ppb. NO$_2$level of occupational environment were 27.87$\pm$7.15 ppb in office, 33.60$\pm$8.64 ppb in platform and 50.13$\pm$13.04 ppb in outdoor. Personal exposure time of subway station workers was constituted as survey results with $7.94\pm$3.00 hours in office, $2.82\pm$1.63 hours in platform and 1 hours in outdoor. With above results, personal $NO_2$exposure distributions on subway station workers in Seoul were estimated with Monte Carlo simulation which uses statistical probabilistic theory on various exposure scenario testing. Some of distributions which did not have any formal patterns were assumed as custom distribution type. Estimated personal occupational $NO_2$exposure using time weighted average (TWA) model was 31.$29\pm$5.57 ppb, which were under Annual Ambient Standard (50ppb) of Korea. Though arithmetic means of measured personal $NO_2$exposure was lower than that of occupational $NO_2$exposure estimated by TWA model, considering probability distribution type simulated, probability distribution of measured personal $NO_2$exposures for 24 hours was over ambient standard with 3.23%, which was higher than those of occupational exposure(0.02%). Further research is needed for reducing these 24 hour $NO_2$personal excess exposures besides occupational exposure on subway station workers in Seoul.

• Journal of KIISE:Computing Practices and Letters
• /
• v.14 no.5
• /
• pp.527-531
• /
• 2008

For statistical modeling, the model parameters are usually estimated by maximizing a probability measure, such as the likelihood or the posterior. In contrast, a variational Bayesian method treats the parameters of a model as probability distributions and computes optimal distributions for them rather than values. It has been shown that this approach effectively avoids the overfitting problem, which is common with other parameter optimization methods. This paper applies a variational Bayesian technique to surface fitting for height field data. Then, we propose point cloud denoising based on the basic surface fitting technique. Validation experiments and further tests with scan data verify the robustness of the proposed method.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.11 no.4
• /
• pp.193-202
• /
• 2008

The objectives of this study are to analyze the spatial site characteristics of existing tombs and the change in the pattern of spatial distributions of tombs over time. The spatial distributions of tombs located in Honam province along the Honam expressway were investigated by interpreting digital aerial photographs taken in two different points of time; 1990 and 2000. According to the results of the study, the tombs newly observed in 2000 photos were located closer to roads and villages than those found in the photos of 1990. This is a finding indicating that the accessibility of tombs has been more important consideration in determining the location of tomb sites. Also found were the gentle slopes of southern aspects to be favored as tomb sites. Based on the data sets of tombs locations and their topographic site characteristics, the probability function of tombs appearance in the study area was derived using the logistic regression analysis technique. As a result, tomb sites were classified as 74.7% by logistic regression. All of six input factors (elevation, slope, aspect, distance from the roads, the town and the stream, respectively) affected the probability of tombs appearance significantly.

• The Korean Journal of Applied Statistics
• /
• v.6 no.2
• /
• pp.341-353
• /
• 1993

As an estimator of the conditional probability of discovering a new species at the next observation after a sample of certain size is taken, the one proposed by Good(1953) has been most widely used. Recently, Clayton and Frees(1987) showed via simulation that their nonparametric maximum likelihood estimator(NPMLE) has smaller MSE than Good's estimator when the population is relatively nonuniform. Lee(1989) proved that their conjecture is asymptotically true for truncated geometric population distributions. One shortcoming of the NPMLE, however, is that it has a considerable amount of negative bias. In this study we proposed a bias-corrected version of the NPMLE for virtually all realistic population distributions. We also showed that it has a smaller asymptotic MSE than Good's extimator except when the population is very uniform. A Monte Carlo simulation was performed for small sample sizes, and the result supports the asymptotic results.

• Journal of Korea Water Resources Association
• /
• v.32 no.2
• /
• pp.197-210
• /
• 1999

This study examines the duration and severity of droughts by the use of stochastic process considerations. The key annual flow statistics are used to estimate the related statistics of drought probability distributions for various combinations of return period and water demand. This study efforts initially focused on analyzing all the nation streamgage records that were judged to meet certain selection criteria, including those of record length, record quality. These analyses resulted in the determination of those annual flow statistics necessary to define the behavior of drought sequences for the selected streams. Using prior research results, the actual or estimated flow statistics are related to the probability distributions of maximum drought events, through the application of the theory of runs. This has resulted in assigning return periods to drought events at gaged locations, and permits an assessment of the probabilities of observed historical drought within the nation.

• Food Science of Animal Resources
• /
• v.39 no.4
• /
• pp.565-575
• /
• 2019

This study evaluated Campylobacter jejuni risk in ground meat products. The C. jejuni prevalence in ground meat products was investigated. To develop the predictive model, survival data of C. jejuni were collected at $4^{\circ}C-30^{\circ}C$ during storage, and the data were fitted using the Weibull model. In addition, the storage temperature and time of ground meat products were investigated during distribution. The consumption amount and frequency of ground meat products were investigated by interviewing 1,500 adults. The prevalence, temperature, time, and consumption data were analyzed by @RISK to generate probabilistic distributions. In 224 samples of ground meat products, there were no C. jejuni-contaminated samples. A scenario with a series of probabilistic distributions, a predictive model and a dose-response model was prepared to calculate the probability of illness, and it showed that the probability of foodborne illness caused by C. jejuni per person per day from ground meat products was $5.68{\times}10^{-10}$, which can be considered low risk.

• International Journal of Naval Architecture and Ocean Engineering
• /
• v.12 no.1
• /
• pp.755-767
• /
• 2020

A significant development has been made on a new fatigue damage model applicable to Gaussian wide band stress response spectra using numerical approximation methods such as data processing, time simulation, and regression analysis. So far, most of the alternative approximate models provide slightly underestimated or overestimated damage results compared with the rain-flow counting distribution. A more reliable approximate model that can minimize the damage differences between exact and approximate solutions is required for the practical design of ships and offshore structures. The present paper provides a detailed description of the development process of a new fatigue damage model. Based on the principle of the Gaussian wide band model, this study aims to develop the best approximate fatigue damage model. To obtain highly accurate damage distributions, this study deals with some prominent research findings, i.e., the moment of rain-flow range distribution MRR(n), the special bandwidth parameter μk, the empirical closed form model consisting of four probability density functions, and the correction factor QC. Sequential prerequisite data processes, such as creation of various stress spectra, extraction of stress time history, and the rain-flow counting stress process, are conducted so that these research findings provide much better results. Through comparison studies, the proposed model shows more reliable and accurate damage distributions, very close to those of the rain-flow counting solution. Several significant achievements and findings obtained from this study are suggested. Further work is needed to apply the new developed model to crack growth prediction under a random stress process in view of the engineering critical assessment of offshore structures. The present developed formulation and procedure also need to be extended to non-Gaussian wide band processes.