• Proceedings of the KSPS conference
• /
• 2003.05a
• /
• pp.123-126
• /
• 2003

We discuss how to reduce the number of inverse matrix and its dimensions requested in MLLR framework for speaker adaptation. To find a smaller set of variables with less redundancy, we employ PCA(principal component analysis) and ICA(independent component analysis) that would give as good a representation as possible. The amount of additional computation when PCA or ICA is applied is as small as it can be disregarded. The dimension of HMM parameters is reduced to about 1/3 ~ 2/7 dimensions of SI(speaker independent) model parameter with which speech recognition system represents word recognition rate as much as ordinary MLLR framework. If dimension of SI model parameter is n, the amount of computation of inverse matrix in MLLR is proportioned to O($n^4$). So, compared with ordinary MLLR, the amount of total computation requested in speaker adaptation is reduced to about 1/80~1/150.

• Journal of Institute of Control, Robotics and Systems
• /
• v.8 no.11
• /
• pp.976-983
• /
• 2002

We presents a modified variable dimension filter with input estimation for maneuvering target tracking. The conventional variable dimension filter with input estimation(VDIE) consists of the input estimation(IE) technique and the variable dimension(VD) filter. In the VDIE, the IE technique is used for estimation of a maneuver onset time and its magnitude in the least square sense. The detection of the maneuver is declared according to the estimated magnitude of the maneuver. The VD filter structure is applied for the adaptation to the maneuver of the target after compensating the filter parameter with respect to the estimated maneuver when the detection of the maneuver is declared. The VDIE is known as one of the best maneuvering target tracking filter based on a single filter. However, it requires too much computational burden since the IE technique is performed at every sampling instance and thus it is computationally inefficient. We propose another variable dimension filter with input estimation named 'Modified VDIE' which combines VD filter with If technique. Modified VDIE has less computational load than the original one by separating maneuver detection and input estimation. Simulation results show that the proposed VDIE is more efficient and outperforms in terms of computational load.

• Korean Journal of Air-Conditioning and Refrigeration Engineering
• /
• v.17 no.7
• /
• pp.611-619
• /
• 2005

In a previous work of the authors, the heat and mass transfer in a desiccant rotor was analyzed theoretically through linearization assumptions and four dimensionless parameter groups dominating the dehumidification process were arranged. In this work is verified whether the four dimensionless parameters also play the dominant roles in more realistic situations where the nonlinear factors affect the heat and mass transfer. The results show that the dehumidification characteristics are closely similar to each other as long as the four dimensionless parameters have the same set of values while the rotor configurations and/or the operation conditions are different from each other. The four dimensionless parameters are $\Psi,\;\chi,\;\sigma$ and N, where $\Psi$ implies the average gradient of relative humidity lines in the psychrometric chart, $\chi$ the heat capacity of the rotor and $\sigma$ the sorption capacity of the rotor, and N implies the number of transfer unit.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2000.12a
• /
• pp.181-184
• /
• 2000

The new method of feature extraction is proposed, considering the statistic feature of human voice, unlike the conventional methods of voice extraction. PCA(principal Component Analysis) is applied to this new method. PCA removes the repeating of data after finding the axis direction which has the greatest variance in input dimension. Then the new method is applied to real voice recognition to assess performance. When results of the number recognition in this paper and the conventional Mel-Cepstrum of voice feature parameter are compared, there is 0.5% difference of recognition rate. Better recognition rate is expected than word or sentence recognition in that less convergence time than the conventional method in extracting voice feature. Also, better recognition tate is expected when the optimum vector is used by statistic feature of data.

• Journal of Korean Society for Atmospheric Environment
• /
• v.16 no.5
• /
• pp.539-544
• /
• 2000

A simple analytic model is proposed here to analyze the concentration deficit field caused by a large area of vegetated area. With non-dimensional deposition velocity chosen as small parameter, the regular perturbation method is exploited to derive the mass balance equation and the dynamic equations for the concentration deficit field, Analytic solutions to those equations are obtained in a closed form for several cases of interest, assuming that the concentration field is stationary and the plume can be nicely approximated as Gaussian for a point source. The results suggest that quite a negligible fraction (less than 1%) of the gaseous air pollutants emitted into the air is removed by the vegetated area of which width is 4 km in wind-wise direction, the typical dimension of the Restricted Development Zones around the metropolitan regions in South Korea.

• Proceedings of the KSRS Conference
• /
• 2007.10a
• /
• pp.247-249
• /
• 2007

Crown parameters are important roles in tree species identification, because the canopy is the aggregate of all the crowns. However, crown measurements with spaceborne image data have remained more difficult than on aerial photographs since trees show more structural detail at higher resolutions. This recognized problem led to the initiation of the research to determine if high resolution satellite image data could be used to identify and classify single tree species. In this paper, shape parameters derived from pixel-based crown area measurements and texture features derived from GLCM parameters in QuickBird image were tested and compared for individual tree species identification. As expected, initial studies have shown that the crown parameters and the canopy texture parameters provided a differentiating method between coniferous trees and broad-leaved trees within the compartment(less than forest stand) for single extraction from spaceborne high resolution image.

• Laser Solutions
• /
• v.2 no.2
• /
• pp.8-17
• /
• 1999

This study includes a basic feature of laser surface alloying for enhancing the surface properties of materials. Effects of laser processing parameters such as beam power, beam size, scanning speed on the shape and composition of alloyed layer was simulated in case of moving beam conditions (2-dimensional numerical methods). Simulated results were compared with experiments, in which the plasma coating of 80% Ni ＋ 20% Cr deposited on the SS41 substrate was remelted with CO2 laser with Gaussian energy distribution. Simulation and experiments revealed that the shape (dimension)and composition of laser alloyed layer were strongly dependent upon the process parameters, especially interaction time (travel speed) as compared to beam diameter, beam power and absorptivity. The shape and composition of alloyed layervaried more or less exponentially with parameters.

• Transactions on Electrical and Electronic Materials
• /
• v.9 no.1
• /
• pp.16-19
• /
• 2008

In this study, it was investigated about optical simulation in direct-type backlight design. Direct-type backlight has been used high-brightness backlight such as medical LCD application. The key parameter in designing direct-type backlight was consists of three geometrical dimension such as the distance of two lamps, the gap of lamp and reflection plate and the number of lamps. It has many of variations in optical design and it causes the different properties in backlight system. It shows the best values of above parameters; 26 mm of the distance of two lamps, 4.5 mm of the gap of lamp and reflection plate and 16 lamps. And we produced the specimen as above condition, and acquired good result in backlight such as the value of the brightness is 6423 nit in center of emission area and less than 5 % in brightness uniformity. It shows the effective ways of designing backlight system using optical simulation method for medical LCD application.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2007.06a
• /
• pp.415-415
• /
• 2007

In this study. it was investigated about optical simulation in direct-type backlight design. Direct-type backlight has been used high-brightness backlight such as LCD television application. The key parameter in designing direct-type backlight was consists of three geometrical dimension such as the distance of two lamps. the gap of lamp and reflection plate and the number of lamps. It has many variation in optical design and it causes the different properties in backlight system. It shows the best values of above parameters; 26mm of the distance of two lamps. 4.5mm of the gap of lamp and reflection plate and 16 lamps. And we produced the specimen as above condition. and acquired good result in backlight such as the value of the brightness is 6436 nit in center of emission area and less than 5% in brightness uniformity. It shows the effective ways of designing backlight system using optical simulation method.

• Geomechanics and Engineering
• /
• v.21 no.3
• /
• pp.289-300
• /
• 2020

Estimating the damage induced by an explosion around a blast hole has always been a challenging issue in geotechnical engineering. It is difficult to determine an exact dimension for damage zone since many parameters are involved in the formation of failures, and there are some uncertainties lying in these parameters. Thus, the present study adopted a probabilistic approach towards this problem. First, a reliability model of the problem was established and the failure probability of induced damage was calculated. Then, the corresponding exceedance risk curve was developed indicating the relation between the failure probability and the cracked zone radius. The obtained risk curve indicated that the failure probability drops dramatically by increasing the cracked zone radius so that the probability of exceedance for any crack length greater than 4.5 m is less than 5%. Moreover, the effect of each parameter involved in the probability of failure, including blast hole radius, explosive density, detonation velocity, and tensile strength of the rock, was evaluated by using a sensitivity analysis. Finally, the impact of the decoupling ratio on the reduction of failures was investigated and the location of its maximum influence was demonstrated around the blast point.