• International journal of advanced smart convergence
• /
• v.7 no.2
• /
• pp.33-37
• /
• 2018

In this study, we propose an iterative clustering algorithm that automatically clusters a set of voice signal data without a label into an optimal number of clusters and generates hmm model for each cluster. In the clustering process, the likelihood calculations of the clusters are performed using iterative hmm learning and testing while varying the number of clusters for given data, and the maximum likelihood estimation method is used to determine the optimal number of clusters. We tested the effectiveness of this clustering algorithm on a small-vocabulary digit clustering task by mapping the unsupervised decoded output of the optimal cluster to the ground-truth transcription, we found out that they were highly correlated.

• KIEE International Transactions on Electrophysics and Applications
• /
• v.11C no.4
• /
• pp.99-106
• /
• 2001

An overview of Silicon Carbide (SiC) Static Induction Transistor (SIT) development is presented. Basic conduction mechanisms are introduced and discussed, including ohmic, exponential, and space charge limited conduction (SCLC) mechanisms. Additionally, the impact of velocity saturation and temperature effects on SCLC are reviewed. The small-signal model, breakdown voltage, power density, and different gate structures are also discussed, before a final review of published SiC SIT results. Published S-band (3-4 GHz) results include 9.5 dB of gain and output power of 120 W, and L-band (1.3 GHz) results include 400 W output power, 7.7 dB of gain, and power density of 16.7 W/cm.

• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 2004.03a
• /
• pp.287-293
• /
• 2004

A numerical simulation is presented for investigating the effects of pressure ratio of $D_2$ injector to supersonic nozzle on the population inversion in the DF chemical laser cavity, while a lasing concurrently takes place. The laser beam is generated between the mirrors in the cavity and it is important to obtain stronger population inversion and more uniform distribution of the excited molecules in the laser cavity in order to produce high power laser beam with good quality. In this study, these phenomena are investigated by means of analyzing the distributions of the DF excited molecules and the F atom used as an oxidant, while simultaneously estimating the maximum small signal and saturated gains and power in the DF chemical laser cavity. For the numerical solution, an 11-species (including DF molecules in various excited states of energies), 32-step chemistry model is adopted for the chemical reaction of the DF chemical laser system. The results are discussed by comparison with two $D_2$injector pressure cases; 192 torr and 388.64 torr. Major results reveal that in the resonator, stronger population inversions occur in the all transitions except DF(1)-DF(0), when the $D_2$injection pressure is lower. But, the higher $D_2$injection pressure provides a favorable condition for DF(1)-DF(0) transition to generate the higher power laser beam. In other words, as the pressure of $D_2$injector increases, the maximum small signal gain in the $V_{1-0}$ transition, which is in charge of generating most of laser power, becomes higher. Therefore, the total laser beam power becomes higher.r.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.11 no.1
• /
• pp.75-82
• /
• 2007

In advanced countries, a gas turbine engine is developed to use in aircraft, vessels, and target weapons. Our nation also passed the level of producing engine components and now, we are developing small-sized gas turbine engine. The most important point of the gas turbine engine, the engine control technique, is evaded by the advanced nations. This document contains the research about the development of the gas turbine engine simulator. The simulator presented in this document has a mathematical engine model based on a capacity data of the gas turbine engine to advance the engine simulator. Through this process, it eases the development of the gas turbine engine control algorithm and helps to check the engine controller function. In this simulator, the engine sensor signal conversion board is designed, so the engine model shows like a real sensor signal during the simulation. Also, this paper contrasts the actual engine test with the simulation results to verify the performance.

• Journal of Korean Society of Transportation
• /
• v.22 no.7 s.78
• /
• pp.161-167
• /
• 2004

While most or fixed-time control systems such as UTCS produce the signal timing plans that either maximizing bandwidth or minimizing a disutility index of delay and stops, cannot consider the fluctuation of traffic flow. One category of the traffic-response control systems, which make small changes on a predefined signal plan such as SCOOT, cannot be easily modified for feedback real-time control schemes based on observation of variables other than traffic flow. The other category, which decide to whether switch the traffic lights or not at each step of time as in PRODYN, does not adequately consider the relations between traffic flows and traffic lights at each step of time. In this paper we present a complete formulation that adequately consider the relations between traffic flows and traffic lights at each step of time. The formulation is a binary mixed integer linear programing (BMILP) that obtain traffic lights at each step for minimizing delay. Since numarical examples for application of the proposed model illustrated that the model adequately produced dynamic traffic signal plans minimizing delay at each step, the model may be expected to contribute to advanced transportation management systems (ATMS) for dynamic traffic signal control.

• Proceedings of the Korea Database Society Conference
• /
• 1999.06a
• /
• pp.175-186
• /
• 1999

Detecting the features of significant patterns from their own historical data is so much crucial to good performance specially in time-series forecasting. Recently, a new data filtering method (or multi-scale decomposition) such as wavelet analysis is considered more useful for handling the time-series that contain strong quasi-cyclical components than other methods. The reason is that wavelet analysis theoretically makes much better local information according to different time intervals from the filtered data. Wavelets can process information effectively at different scales. This implies inherent support fer multiresolution analysis, which correlates with time series that exhibit self-similar behavior across different time scales. The specific local properties of wavelets can for example be particularly useful to describe signals with sharp spiky, discontinuous or fractal structure in financial markets based on chaos theory and also allows the removal of noise-dependent high frequencies, while conserving the signal bearing high frequency terms of the signal. To date, the existing studies related to wavelet analysis are increasingly being applied to many different fields. In this study, we focus on several wavelet thresholding criteria or techniques to support multi-signal decomposition methods for financial time series forecasting and apply to forecast Korean Won / U.S. Dollar currency market as a case study. One of the most important problems that has to be solved with the application of the filtering is the correct choice of the filter types and the filter parameters. If the threshold is too small or too large then the wavelet shrinkage estimator will tend to overfit or underfit the data. It is often selected arbitrarily or by adopting a certain theoretical or statistical criteria. Recently, new and versatile techniques have been introduced related to that problem. Our study is to analyze thresholding or filtering methods based on wavelet analysis that use multi-signal decomposition algorithms within the neural network architectures specially in complex financial markets. Secondly, through the comparison with different filtering techniques' results we introduce the present different filtering criteria of wavelet analysis to support the neural network learning optimization and analyze the critical issues related to the optimal filter design problems in wavelet analysis. That is, those issues include finding the optimal filter parameter to extract significant input features for the forecasting model. Finally, from existing theory or experimental viewpoint concerning the criteria of wavelets thresholding parameters we propose the design of the optimal wavelet for representing a given signal useful in forecasting models, specially a well known neural network models.

• Proceedings of the Korea Inteligent Information System Society Conference
• /
• 1999.03a
• /
• pp.175-186
• /
• 1999

Detecting the features of significant patterns from their own historical data is so much crucial to good performance specially in time-series forecasting. Recently, a new data filtering method (or multi-scale decomposition) such as wavelet analysis is considered more useful for handling the time-series that contain strong quasi-cyclical components than other methods. The reason is that wavelet analysis theoretically makes much better local information according to different time intervals from the filtered data. Wavelets can process information effectively at different scales. This implies inherent support for multiresolution analysis, which correlates with time series that exhibit self-similar behavior across different time scales. The specific local properties of wavelets can for example be particularly useful to describe signals with sharp spiky, discontinuous or fractal structure in financial markets based on chaos theory and also allows the removal of noise-dependent high frequencies, while conserving the signal bearing high frequency terms of the signal. To data, the existing studies related to wavelet analysis are increasingly being applied to many different fields. In this study, we focus on several wavelet thresholding criteria or techniques to support multi-signal decomposition methods for financial time series forecasting and apply to forecast Korean Won / U.S. Dollar currency market as a case study. One of the most important problems that has to be solved with the application of the filtering is the correct choice of the filter types and the filter parameters. If the threshold is too small or too large then the wavelet shrinkage estimator will tend to overfit or underfit the data. It is often selected arbitrarily or by adopting a certain theoretical or statistical criteria. Recently, new and versatile techniques have been introduced related to that problem. Our study is to analyze thresholding or filtering methods based on wavelet analysis that use multi-signal decomposition algorithms within the neural network architectures specially in complex financial markets. Secondly, through the comparison with different filtering techniques results we introduce the present different filtering criteria of wavelet analysis to support the neural network learning optimization and analyze the critical issues related to the optimal filter design problems in wavelet analysis. That is, those issues include finding the optimal filter parameter to extract significant input features for the forecasting model. Finally, from existing theory or experimental viewpoint concerning the criteria of wavelets thresholding parameters we propose the design of the optimal wavelet for representing a given signal useful in forecasting models, specially a well known neural network models.

• Journal of Advanced Marine Engineering and Technology
• /
• v.13 no.1
• /
• pp.57-68
• /
• 1989

This paper treats the analytical and experimental studies on the improvement of commutating circuit for the speed control of DC motor. A simple circuit composed of R, L and C elements is proposed here for switching off power SCR carrying the load current. The real important in this chopper circuit is to determine the reasonable values of commutating circuit constants. In this paper, the reasonable values of the commutating circuit constants are basically determined on a view point of commutating performances in the given circuit model and must satisfy the following conditions. The first, the peak commutating current should be larger than the anticipated maximum load current. The second, the circuit turn-off time (tc) must be longer than the SCR turn-off time (tq). The third, the resistor should be enough large to permit the current to be neglected in the analysis of the commutation circuit, as well as be enough small to permit to charge the capacity voltage (Ec) to the half the value of source voltage (E) before the next communication cycle is initiated. The last, the period of chopping signal must be the least possible multiple of the damping vibration period of commutating circuit. The improved chopper circuit used in the experiment under unloaded condition was composed to meet the reasonable conditions mentioned above, and a successful commuting performance was achieved without failure. Several types of microprocessor having a different value of CPU speed individually have been applied to the experiment under the loaded conditions. Also it shows that the faster the speed of CPU is, the more stable the commutation turns out.

• Journal of The Korean Astronomical Society
• /
• v.42 no.6
• /
• pp.145-153
• /
• 2009

We present measurements of diffuse interstellar $H_2$ absorption lines in the continuum spectra of 10 early-type stars. The data were observed with the Berkeley Extreme and Far-Ultraviolet Spectrometer (BEFS) of the ORFEUS telescope on board the ORFEUS-SPAS I and II space-shuttle missions in 1993 and 1996, respectively. The spectra extend from the interstellar cutoff at 912 $\AA$ to about 1200 $\AA$ with a resolution of ~ 3000 and statistical signal-to-noise ratios between 10 and 65. Adopting Doppler broadening velocities from high-resolution optical observations, we obtain the $H_2$ column densities of rotational levels J" = 0 through 5 for each line of sight. The kinetic temperatures derived from J" = 0 and 1 states show a small variation around the mean value of 80 K, except for the component toward HD 219188, which has a temperature of 211 K. Based on a synthetic interstellar cloud model described in our previous work, we derive the incident UV intensity IUV and the hydrogen density $n_H$ of the observed components to be -0.4 $\leq$ log $I_{UV}\leq2.2$ and $6.3{\leq}n_H2500cm^{-3}$, respectively.

• The Korean Journal of Nuclear Medicine Technology
• /
• v.27 no.2
• /
• pp.111-127
• /
• 2023

Purpose: Gallium-68 (⁶⁸Ga) is increasingly used in nuclear medicine imaging for various conditions such as lymphoma and neuroendocrine tumors by labeling tracers like Prostate Specific Membrane Antigen (PSMA) and DOTA-TOC. However, compared to Fluorine-18 (¹⁸F) used in conventional nuclear medicine imaging, ⁶⁸Ga has lower spatial resolution and relatively higher Signal to Background Ratio (SBR). Therefore, this study aimed to investigate the optimized parameters and reconstruction methods for PET/CT imaging using the ⁶⁸Ga radiotracer through model-based image evaluation. Materials and Methods: Based on clinical images of ⁶⁸Ga-PSMA PET/CT, a NEMA/IEC 2008 PET phantom model was prepared with a Hot vs Background (H/B) ratio of 10:1. Images were acquired for 9 minutes in list mode using DMIDR (GE, Milwaukee WI, USA). Subsequently, reconstructions were performed for 1 to 8 minutes using OS-EM (Ordered Subset Expectation Maximization) + TOF (Time of Flight) + Sharp IR (VPFX-S), and BSREM (Block Sequential Regularized Expectation Maximization) + TOF + Sharp IR (QCFX-S-400), followed by comparative evaluation. Based on the previous experimental results, images were reconstructed for BSREM + TOF + Sharp IR / 2 minutes (QCFX-S-2min) with varying β-strength values from 100 to 700. The image quality was evaluated using AMIDE (freeware, Ver.1.0.1) and Advanced Workstation (GE, USA). Results: Images reconstructed with QCFX-S-400 showed relatively higher values for SNR (Signal to Noise Ratio), CNR (Contrast to Noise Ratio), count, RC (Recovery Coefficient), and SUV (Standardized Uptake Value) compared to VPFX-S. SNR, CNR, and SUV exhibited the highest values at 2 minutes/bed acquisition time. RC showed the highest values for a 10 mm sphere at 2 minutes/bed acquisition time. For small spheres of 10 mm and 13 mm, an inverse relationship between β-strength increase and count was observed. SNR and CNR peaked at β-strength 400 and then decreased, while SUV and RC exhibited a normal distribution based on sphere size for β-strength values of 400 and above. Conclusion: Based on the experiments, PET/CT imaging using the ⁶⁸Ga radiotracer yielded the most favorable quantitative and qualitative results with a 2 minutes/bed acquisition time and BSREM reconstruction, particularly when applying β-strength 400. The application of BSREM can enhance accurate quantification and image quality in ⁶⁸Ga PET/CT imaging, and an optimization process tailored to each institution's imaging objectives appears necessary.