• 대한음성학회:학술대회논문집
• /
• 대한음성학회 2006년도 춘계 학술대회 발표논문집
• /
• pp.89-93
• /
• 2006

This paper presents ETRI broadcast news speech recognition system. There are two major issues on the broadcast news speech recognition: 1) real-time processing and 2) out-of-vocabulary handling. For real-time processing, we devised the dual decoder architecture. The input speech signal is segmented based on the long-pause between utterances, and each decoder processes the speech segment alternatively. One decoder can start to recognize the current speech segment without waiting for the other decoder to recognize the previous speech segment completely. Thus, the processing delay is not accumulated. For out-of-vocabulary handling, we updated both the vocabulary and the language model, based on the recent news articles on the internet. By updating the language model as well as the vocabulary, we can improve the performance up to 17.2% ERR.

• 경영과학
• /
• 제32권3호
• /
• pp.91-103
• /
• 2015

This study recognizes that there is a correlation between the movement of the financial market and the sentimental changes of the public participating directly or indirectly in the market, and applies the relationship to investment strategies for stock market. The concerns that market participants have about the economy can be transformed to the search terms that internet users query on search engines, and search volume of a specific term over time can be understood as the economic trend of big data. Under the hypothesis that the time when the economic concerns start increasing precedes the decline in the stock market price and vice versa, this study proposes three investment strategies using casuality between price of domestic stock market and search volume from Naver trends, and verifies the hypothesis. The computational results illustrate the potential that combining extensive behavioral data sets offers for a better understanding of collective human behavior in domestic stock market.

• 한국경영과학회지
• /
• 제27권2호
• /
• pp.97-109
• /
• 2002

Manufacturing environments in the real world are subject to many sources of change and uncertainty, such as new job releases, job cancellations, a chance in the processing time or start time of some operation. Thus, the realistic scheduling method should Properly reflect these dynamic environment. Based on the release times of jobs, JSSP (Job Shoe Scheduling Problem) can be classified as static and dynamic scheduling problem. In this research, we mainly consider the dynamic JSSP with continually arriving jobs. The goal of this research is to develop an efficient scheduling method based on GA (Genetic Algorithm) to address dynamic JSSP. we designed scheduling method based on SGA (Sing1e Genetic Algorithm) and PGA (Parallel Genetic Algorithm) The scheduling method based on GA is extended to address dynamic JSSP. Then, This algorithms are tested for scheduling and rescheduling in dynamic JSSP. The results is compared with dispatching rule. In comparison to dispatching rule, the GA approach produces better scheduling performance.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2000년도 제15차 학술회의논문집
• /
• pp.208-208
• /
• 2000

In realistic control systems, the nonlinear salutation attributes of the control actuator due to physical limitations should be taken into account. This nonlinear saturation of actuators may cause not only deterioration of the control performance but also a large overshoot during start-up and shut-down. As the overshoot increases, the system may become oscillatory unstable. In this paper, the supervisor implementation which guarantees good performance lot saturation operation and prevents reset wind-up is presented. Moreover, the sufficient conditions of the stability for saturated systems using supervisory control with a dynamic controller are provided in the discrete-time domain. A numerical example is illustrated to depict the efficiency of supervisory control for a typical saturated production-distribution system controlled by a discrete-time dynamic controller and to validate basic results by simulation.

• 로봇학회논문지
• /
• 제2권2호
• /
• pp.183-190
• /
• 2007

The most important thing for navigation of a mobile robot is to find the most suitable path and avoid the obstacles in the static and dynamic environment. This paper presents a method to search the optimal path in start space extended to time domain with considering a velocity and a direction of moving obstacles. A modified version of $A^*$ algorithm has been applied for path planning in this work and proposed a method of path search to avoid a collision with moving obstacle in space-tim domain with a velocity and an orientation of obstacles. The velocity and the direction for moving obstacle are assumed as linear form. The simulation result shows that a mobile robot navigates safely among moving obstacles of constant linear velocity. This work can be applied for not only a moving robot but also a legged humanoid robot and all fields where the path planning is required.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제8권11호
• /
• pp.3638-3654
• /
• 2014

Cognitive Radio (CR) users need to sense the environment or channel at regular time interval for sharing the spectrum band of the primary users (PUs). Once find the spectrum idle, CR users start their transmission through it. Even while transmitting, they need to continue the sensing process so that they can leave the spectrum immediately whenever find a PU wanting to use the band. Therefore, detecting PUs is one of the main functions of cognitive radio before transmission and higher the detection probability ensures better protection to the primary users. However, it is not possible to attain a high detection probability (or a low miss detection probability) and low false alarm probability simultaneously as there is a tradeoff between false alarm probability ($P_{fa}$) and the probability of detection ($P_d$). In this paper, the author has provided a comprehensive study on different sensing techniques and discussed their advantages and disadvantages. Moreover, it is expected that, with this article, readers can have a through understanding of sensing techniques in CR and the current research trends in this area.

• 한국전기전자재료학회논문지
• /
• 제13권4호
• /
• pp.337-342
• /
• 2000

We fabricated resistive superconducting fault current limiters based on YB $a_{2}$/C $u_{3}$/ $O_{7}$ thin films and investigated their quench propagation characteristics. The YB $a_{2}$/C $u_{3}$/ $O_{7}$ films was coated with a gold layer and patterned into 1 mm wide meander lines by photolithography. The quench was concluded to start locally and propagates until completed. The quench propagation characteristics were explained based on the heat transfer within the film as well as between the film and the surrounding liquid nitrogen. The quench completion time depended strongly on potential fault current amplitude and not significantly on fault angle which indicates that the quench propagation speed is affected more by heat dissipation rate than by fault current increase rate. The quench completion time was 1 msec at the fault current of 65 $A_{peak/{\ak}}$.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2006년도 제37회 하계학술대회 논문집 B
• /
• pp.1099-1100
• /
• 2006

The running of the electric train are done by powering, breaking and coasting. Powering to start and accelerate the rolling stock means driving the three-phase electric induction motors controlled by VVVF inverters which are connected to overhead power line directly(DC) or indirectly(AC, DC/AC). Breaking is achieved by blending control which is the proportional combination between regenerative breaking of VVVF inverter and air pressure control of breaking operating unit(BOU). Therefore, Data obtained during two operation are very important items to evaluate the running performance of RS. This paper has investigated in real time data acquired from VVVF inverter and measured by pressure sensor directly connected to air breaking line(motor car and trailer car). By analyzing data in the region of time and frequency, fundamental methods for eavaluating the running performance of RS quantitatively and objectively have been suggested.

• 한국분무공학회지
• /
• 제7권2호
• /
• pp.22-30
• /
• 2002

A number of droplet breakup models have been developed to predict the diesel spray. The capabilities of droplet deformation and breakup models such as TAB, ETAB, DDB and APTAB models are evaluated in modeling the non-evaporating diesel sprays injected into atmosphere. New methods are also suggested that take into account the non- spherical shape of droplets and the reduced drag force by the presence of neighbouring droplets. The KIVA calculations with standard ETAB, DDB, and APTAB models predict well the spray tip penetrations of the experiment, but overestimate the Sauter mean Diameter(SMD) of droplets. The calculation with non spherical droplets injected from the nozzle shows very similar results to the calculation with spherical droplets. The drag coefficient which is linearly increased with the time after start of injection during the breakup time gives the smaller SMD that agrees well with the experimental result.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 2009년도 춘계학술대회 논문집
• /
• pp.360-364
• /
• 2009

To verify the suitability of the Railroad safety requirements analysis, the actions of the function which it follows in a given operation condition are composed of the rule model or the state model. And it test must be possible. The subject of condition transfer in compliance with the transfer condition which is an event of system it will be able to accomplish the operation of system. However, tasks allocated to the actor are the duty which is added in organization. This time, the start of duty task is started in compliance with an external or internal condition. If there are 9 tasks which have one objective, the relationship among the previous and the current internal tasks and next tasks which will given is able to show in logical model. Thus when planning an organization, the integral part of class must include Task, Role and Position. And this time we make a series of outputs of examination of organizing model and examples to meet the capable skill.