• Speech Sciences
• /
• v.9 no.1
• /
• pp.39-47
• /
• 2002

This paper proposes an effective method to remedy the acoustic modeling problem inherent in the usual log-normal Parallel Model Composition intended for achieving robust speech recognition. In particular, the Gaussian kernels under the prescribed log-normal PMC cannot sufficiently express the corrupted speech distributions. The proposed scheme corrects this deficiency by judiciously selecting the 'fairly' corrupted component and by re-estimating it as a mixture of two distributions using data-driven PMC. As a result, some components become merged while equal number of components split. The determination for splitting or merging is achieved by means of measuring the similarity of the corrupted speech model to those of the clean model and the noise model. The experimental results indicate that the suggested algorithm is effective in representing the corrupted speech distributions and attains consistent improvement over various SNR and noise cases.

• Journal of ILASS-Korea
• /
• v.23 no.4
• /
• pp.169-174
• /
• 2018

In this work, optimization of blade shape for the improvement of mixture formation and vortex of intake port was performed by numerically, and the combustion performance of CI engine with optimized blade shape was investigated. To achieve this, 3 types of blade shape were studied under the different air flow mass conditions and the numerical results were investigated in terms of humidification water, moisture concentration, and velocity distributions. Evaporated liquid mass was also compared under various test conditions to reveal the turbulent intensity in an intake port. It was observed that the optimized blade shape can improve the humidification water, moisture concentration, and velocity distributions of intake port inside. The evaporated liquid mass was also increased under the conditions with blade. Especially, low NOx emissions was observed with optimized blade condition.

• Journal of Advanced Marine Engineering and Technology
• /
• v.27 no.2
• /
• pp.260-271
• /
• 2003

For the Purpose of understanding the mixing phenomena of a GDI(Gasoline Direct Injection) engine, the spray behaviors and fuel distributions were investigated in a single cylinder transparent GDI engine. The experimental engine is a swirl type GDI engine with a SCV(Swirl Control Valve). PLIF(Planar Laser Induced Fluorescence) system with KrF Excimer laser was used for the measurement of the fuel distributions. The effects of SCV opening angles and the injector specifications on the fuel distribution characteristics were investigated. As a result, it was found that the SCV opening angle had a great effect on the fuel distributions in the late stage of compression process by changing the flow fields in the combustion chamber.

• Communications for Statistical Applications and Methods
• /
• v.7 no.1
• /
• pp.129-140
• /
• 2000

Wavelets constitute a new orthogonal system which has direct application in density estimation. We introduce a brief wavelet density estimation and summarize some asymptotic results. An application to mixture normal distributions is implemented with S-Plus.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.40 no.1
• /
• pp.1-9
• /
• 1991

This paper describes a new method of calculating expected energy generation and loss of load probability (L.O.L.P) for electric power system operation and expansion planning. The method represents an equivalent load duration curve (E.L.D.C) as a mixture of cumulants approximation (M.O.N.A). By regarding a load distribution as many normal distributions-rather than one normal distribution-and representing each of them in terms of Gram-Charlier expansion, we could improve the accuracy of results. We developed an algorithm which automatically determines the number of distribution and demarcation points. In modeling of a supply system, we made subsets of generators according to the number of generator outage: since the calculation of each subset's moment needs to be processed rapidly, we further developed specific recursive formulae. The method is applied to the test systems and the results are compared with those of cumulant, M.O.N.A. and Booth-Baleriaux method. It is verified that the M.O.C.A. method is faster and more accure than any other method.

• 한국연소학회:학술대회논문집
• /
• 2013.06a
• /
• pp.9-10
• /
• 2013

This study reports the results of an experimental investigation of emission and temperatures from the syngas-air premixed flame with a various mixture composition in the region of large equivalence ratios. The effects of hydrogen contents and equivalence ratios on the flame velocity, which reported before, and emission of syngas fuel are examined. In this study, representative syngas mixture compositions ($H_2:CO$) such as $H_2:CO=10:90$, 25:75, 50:50 and 75:25 and equivalence ratios from 0.5 to 5.0 have been conducted. The emissions of syngas fuel were measured by the high precision analyzer with enclosure configuration and the adiabatic temperatures are calculated by used Chemkin basis. The NOx emission level is coincided relatively well with the adiabatic temperature distributions in lean mixture conditions, but for rich mixture conditions NOx level was also increased again even though the adiabatic temperature decreases. Such an increasing characteristics in rich mixture conditions is coincided well with the tendency that rather the flue gas temperature increases.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.15 no.5
• /
• pp.86-96
• /
• 2007

A direct-injection stratified-charge(DISC) engine has been considered as a promising alternative in spite of high unburned hydrocarbon emission levels during light load operation. In this paper investigation is made to characterize formation and combustion processes of stratified mixture charge in a simple constant volume combustion chamber. Both experimental and numerical analyses are performed for fluid and combustion characteristics with 3 different induction types for rich, homogeneous and lean mixture conditions. The commercial code FIRE is applied to the turbulent combustion process in terms of measured and calculated pressure traces and calculated distributions of mean temperature, OH radical and reaction rate. It turns out that the highest combustion rate occurs for the rich state condition at the spark ignition location due to existence of stoichiometric mixture and timing.

• International Journal of Automotive Technology
• /
• v.6 no.6
• /
• pp.579-584
• /
• 2005

Unsteady state free natural gas jets injected from several types of injectors were numerically simulated. Simulations showed good agreements with the schlieren experimental results. Moreover, injections of natural gas in intake manifolds of a single-valve engine and a double-valve engine were predicted as well. Predictions revealed that large volumetric injections of natural gas in intake manifolds led to strong impingement of natural gas with the intake valves, which as a result, gave rise to pronounced backward reflection of natural gas towards the inlets of intake manifolds, together with significant increase in pressure in intake manifold. Based on our simulations, we speculated that for engines with short intake manifolds, reflections of the mixture of natural gas and air were likely to approach the inlets of intake manifolds and subsequently be inbreathed into other cylinders, resulting in non-uniform mixture distributions between the cylinders. For engines with long intake manifolds, inasmuch as the degrees of intake interferences between the cylinders were not identical in light of the ignition sequences, non-uniform intake charge distributions between the cylinders would occur.

• Journal of the Korean Society of Manufacturing Process Engineers
• /
• v.18 no.5
• /
• pp.60-65
• /
• 2019

Since nanofluids (NFs), which are a mixture of a small amount of nanoparticles and a bulk liquid solvent, were first proposed by Stephen Choi at the Argonne National Lab in 1995, they have been considered for use in many technical studies of power cooling systems and their practical application due to their high thermal conductivity and heat transfer coefficients compared to conventional coolants. Although nanofluids are a well-known form of engineering fluid that show great promise for use in future cooling systems, their underlying physics as demonstrated in experiments remain unclear. One proven method of determining the heat transfer performance of nanofluids is measuring the concentration of nanoparticles in a mixture. However, it is experimentally inefficient to build testbeds to systematically observe particle distributions on a nanoscale. In this paper, we demonstrate the distribution of nanoparticles under a temperature gradient in a solution using molecular dynamics simulations. First, temperature profiles based on substrate temperature are introduced. Following this, the radial pair distribution functions of pairs of nanoparticles, solvents, and substrates are calculated. Finally, the distribution of nanoparticles in different heating regions is determined.

• The Korean Journal of Applied Statistics
• /
• v.13 no.2
• /
• pp.343-352
• /
• 2000

In this study. the author proposes a nonhierarchical clustering method. called the "Double K-Means Clustering", which performs clustering of multivariate observations with the following algorithm: Step I: Carry out the ordinary K-means clmitering and obtain k temporary clusters with sizes $n_1$,... , $n_k$, centroids $c_$1,..., $c_k$ and pooled covariance matrix S. $\bullet$ Step II-I: Allocate the observation x, to the cluster F if it satisfies ..... where N is the total number of observations, for -i = 1, . ,N. $\bullet$ Step II-2: Update cluster sizes $n_1$,... , $n_k$, centroids $c_$1,..., $c_k$ and pooled covariance matrix S. $\bullet$ Step II-3: Repeat Steps II-I and II-2 until the change becomes negligible. The double K-means clustering is nearly "optimal" under the mixture of k multivariate normal distributions with the common covariance matrix. Also, it is nearly affine invariant, with the data-analytic implication that variable standardizations are not that required. The method is numerically demonstrated on Fisher's iris data.