• The KIPS Transactions:PartB
• /
• v.8B no.6
• /
• pp.699-704
• /
• 2001

ABSTRACT In this paper, a new hybrid modular recurrent TDNN (time-delay neural network)-HMM (hidden Markov model) architecture for speech recognition has been studied. In TDNN, the recognition rate could be increased if the signal window is extended. To obtain this effect in the neural network, a high-level memory generated through a feedback within the first hidden layer of the neural network unit has been used. To increase the ability to deal with the temporal structure of phonemic features, the input layer of the network has been divided into multiple states in time sequence and has feature detector for each states. To expand the network from small recognition task to the full speech recognition system, modular construction method has been also used. Furthermore, the neural network and HMM are integrated by feeding output vectors from the neural network to HMM, and a new parameter smoothing method which can be applied to this hybrid system has been suggested.

• Journal of Advanced Navigation Technology
• /
• v.24 no.6
• /
• pp.607-612
• /
• 2020

In this paper, we deal with the stability of linear discrete systems with time-varying delays and unstructured uncertainty. Stability conditions are derived based on Lyapunov stability theory, and can include the effect of uncertainty. The unstructured uncertainty in the papaer which can not be figured out its exact characteristics and only can be expreesed by its magnitude is considered. Compared with the previous results on the stability, the new results can expand the applicable systems and alleviate the stability conditions which are more effective and powerful. The proposed stability condition is expressed in the form of an simple inequality, and includes the both effects of the uncertainties and time-varying delay. We present the results comparing the new stability condition with the existing results, and verify the effectiveness and the superiority of the proposed results through numerical example.

• Journal of Physiology & Pathology in Korean Medicine
• /
• v.16 no.5
• /
• pp.1075-1078
• /
• 2002

Objective: The management for the stroke should ,given as soon as possible to be effect. But Patients with stroke symptoms commonly delay many hours before seeking medical attention. We evaluated the factors which are related to the time of hospital arrival after acute stroke. Method: Data were obtained from 317 patients admitted to our hospital within 72 hours of stroke onset. We assessed demographic variables, stoke subtype. referral routes. history of previous stroke, level of consciousness, distance from the place where stroke occurred to hospital, and the time interval between onset of stroke and arrival at the hospital. Results: Mean patient age was 65.99±9.57 years. The mean time interval between onset of stroke and hospital arrival was 17.26±18.69 hours and 128 (40.38%) patients arrived within 6 hours. The patients whoes stoke subtype was infarction, who arrived our hospital by way of other hospital, who had no suffered from previous stroke and who showed no impairement of consciousness was arrived at the hospital late(p<0.05). Conclusion: The majority of patients arrive at the hospital after prolonged delays for multiple reasons, and patients with milder symptoms, for whom treatment might be more effective, were less likely to arrive in time for therapy. Our study suggest that effective education about stroke to the patients and public would be highly necessary.

• Journal of Korean Port Research
• /
• v.13 no.2
• /
• pp.189-198
• /
• 1999

The purpose of this paper is to develop a knowledge-based real-time decision support system to support decision makers for efficient berth operation of Inchon Port. In these days the berth operation problems have been many studied. The berth operation rules differ from port to port and the problem is highly dependent on natural geographical and operational environment of port. In Inchon Port the ship’s entrance into port and departure from port is extremely affected status of dock. In this paper we analyzed some effects of dock a specific character of Inchon Port with a real data of ship’s in Inchon Port. And reconstruct a previous expert’s knowledge of berth allocating problem in Inchon Port. Also the mechanism for the efficient berth operation has been studied by repeatedly dispatching in order to obtain a best effect of berth allocation, with real-time updated information for delay at service time of a specific berth and changing of a working-berth. The system is developed with graphic user interface(GUI) concept using user interactive approach. And this system will be provide decision support maker with an efficient and fast way to berth allocating and reduce wastes of time space and manpower in Inchon Port operation.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.10 no.5
• /
• pp.23-35
• /
• 2011

The pedestrian green time is usually long at wide width intersections. This sometimes causes the increase of delay on the whole intersection because of long cycle length and thus small g/C ratio on some direction. In this paper, to improve these problems, staggered pedestrian crossing was evaluated on the vehicular and pedestrian aspects. The results were gained by using both TRANSYT-7F and VISSIM model. The vehicle control delay of the staggered pedestrian crossing was estimated to be decreasing than that of the general pedestrian crossing by 14.9% to 85.6%. The pedestrian average delay of two pedestrian crossing systems was examined by analytical method and VISSIM. According to the analytical method there was no significant difference between each pedestrian crossing system. The pedestrian delay of staggered pedestrian crossing was from 13.4% to 22.3% than the general pedestrian crossing by VISSIM. In conclusion, the staggered pedestrian crossing was more effective than general pedestrian crossing for both the vehicle and the pedestrian. However this conclusion was resulted from micro simulation where traffic volume condition, v/c, was from 0.8 to 1.1.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.29 no.11
• /
• pp.828-833
• /
• 2018

It is well known that reflected waves and resonance affect phase distortion. In addition, phase delay can be distorted by antenna impedance. In this study, we analyze the phase delay variation caused by the antenna impedance, considering mutual coupling effects. In addition, we confirm the beam steering characteristics. When was -10 dB and -7 dB, the maximum phase delay error was $18.5^{\circ}$ and $26.5^{\circ}$, respectively. The Monte Carlo simulation with an eight-element linear array antenna demonstrated that the RMS error of the beam steering angle ranged from $0.19^{\circ}$ to $0.4^{\circ}$, and the standard deviation ranged from $0.14^{\circ}$ to $0.33^{\circ}$ when the beam steering angle was in the range of $0^{\circ}$ to $30^{\circ}$, with the uniformly distributed phase error of $18.5^{\circ}$ and $26.5^{\circ}$. The side lobe level increased from 0.74 dB to 1.21 dB by the phase error from the theoretical value of -12.8 dB, with a standard deviation of 0.31 dB to 0.51 dB. This is verified by designing an eight-element spiral array antenna.

• Journal of Power Electronics
• /
• v.17 no.2
• /
• pp.389-397
• /
• 2017

A control scheme for stand-alone inverters that utilizes an inductor current observer (ICO) is proposed. The proposed method measures disturbance load currents using a current sensor and it estimates the inductor current using the ICO. The filter parameter mismatch effect is analyzed to confirm the ICO's controllability. The ICO and controllers are designed in a continuous-time domain and transferred to a discrete-time domain with a digital delay. Experimental results demonstrate the effectiveness of the ICO using a 5-kVA single-phase stand-alone inverter prototype. The experimental results demonstrate that the observed current matches the actual current and that the proposed method can archive a less than 2.4% total harmonic distortion (THD) sinusoidal output waveform under nonlinear load conditions.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.34 no.1
• /
• pp.19-28
• /
• 1985

As is well known, the power equations of the N-node system have 2N-1 voltage solutions at most. The vlotage solutions are characterized by the introduction of the mode concept in this paper. There are two mode voltages at one node. One is defined as the (+) mode voltage and the other is defined as the (-) mode one. In this paper, we show that the (-) mode voltage responds to the increase of the power condenser almost adversly to the response of the (+) one. We study how to prevent the voltage collapse in the system with the induction motor load. The critical values of the gain and the time constant in case of the continuous power condenser control, and of the unit power condenser and the closing time delay in case of the discontinuous control for the prevention of the voltage collapse, are calculated. The effect of the composition ratio of the impedance load to the induction moter load on the above critical values are also investigated.

• Explosives and Blasting
• /
• v.16 no.2
• /
• pp.34-46
• /
• 1998

We developed a method to control vibration from some blasting points. This method uses a correlation of vibration waves to decide the most effective delay time to control vibration by interfering vibration waves with each other. We applied this method to the small blasting using mortar blocks and examined the effect on controlling vibration. As the result of the examinations, the maximum vibration velocity by this method caused by five detonators blasting in row of five holes became smaller than that by a detonator blasting in the nearest hole from the measuring point. And the velocity was about one fifth of that of maximum condition in which vibration waves caused by the five detonators arrivved at the same time to the measuring point.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.18 no.2
• /
• pp.49-54
• /
• 2004

In process control, PPl controller including smith predictor was proposed for efficient control method because of the weak performance by the effect of long dead-time. But PPI controller can not guarantee of control performance at real plant because of sensitivity to model mismatch. Thus, In this paper, we presented new control parameter estimation strategy for compensating of model mismatch. The proposed startegy obtained useful result through various cases.