• 대한음성학회지:말소리
• /
• 제56호
• /
• pp.207-223
• /
• 2005

The goal of language model adaptation is to improve the background language model with a relatively small adaptation corpus. This study presents a language model adaptation technique where additional text data for the adaptation do not exist. We propose the information retrieval (IR) technique with N-gram language modeling to collect the adaptation corpus from baseline text data. We also propose to use a dynamic language model interpolation coefficient to combine the background language model and the adapted language model. The interpolation coefficient is estimated from the word hypotheses obtained by segmenting the input speech data reserved for held-out validation data. This allows the final adapted model to improve the performance of the background model consistently The proposed approach reduces the word error rate by $13.6\%$ relative to baseline 4-gram for two-hour broadcast news speech recognition.

• International Journal of Aeronautical and Space Sciences
• /
• 제2권2호
• /
• pp.82-94
• /
• 2001

The purpose of this paper is to analyze the static and dynamic stability-of the unmanned airship under development ; the target airship's over-all length of hull is 50m and the maximum diameter is 12.5m. For the analysis, the dynamic model of an airship was defined and both the nonlinear and linear dynamic equations of motion were derived. Two different configuration models (KA002Y and KA003Y) of the airship were used for the target model of the static stability analysis and the dynamic stability analysis. From the result of analyses, though the airship is unstable in static stability, dynamic characteristics of the airship can provide the stable dynamic stability. All of the results, airship models and dynamic flight equations will be an important basement and basic information for the next step of developing the automatic flight control system(AFCS) and the stability augmentation system(SAS) for the unmanned airship as well as for the stratospheric airship in the future.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2002년도 추계학술대회논문집
• /
• pp.68-73
• /
• 2002

It is necessary first to understand dynamic characteristics of bus full BIW assembly for fatigue endurance analysis. FE model has been used usually for analyzing the dynamic behavior of structures. A lot of experience and effort, however, is necessary to make the credible FE model. Experimental modal analysis of structures has been performed to verify the credibility of initial FE model and to update the model. In this work, experimental modal analysis was performed to understand dynamic characteristics of bus full BIW assembly in free-free boundary condition and the result was used to verify the initial FE model. In addition, some practical techniques, which were used in this experiment, were mentioned.

• Journal of Advanced Marine Engineering and Technology
• /
• 제30권3호
• /
• pp.369-375
• /
• 2006

Large eddy simulation has been applied to simulate turbulent flow around a cubic obstacle mounted on a channel surface for a Reynolds number of 40000(based on the incoming bulk velocity and the obstacle height) using a Smagorinsky model and a Lagrangian dynamic model. In order to develop the LES to the practical engineering application, the effect of upwind scheme, turbulent sub-grid scale model were investigated. The computed velocities. turbulence quantifies, separation and reattachment length were evaluated by compared with the previous experimental results.

• 한국멀티미디어학회논문지
• /
• 제11권6호
• /
• pp.807-815
• /
• 2008

A simple efficient method of spotting and recognizing hand gestures in video is presented using a network of hidden Markov models and dynamic programming search algorithm. The description starts from designing a set of isolated trajectory models which are stochastic and robust enough to characterize highly variable patterns like human motion, handwriting, and speech. Those models are interconnected to form a single big network termed a spotting network or a spotter that models a continuous stream of gestures and non-gestures as well. The inference over the model is based on dynamic programming. The proposed model is highly efficient and can readily be extended to a variety of recurrent pattern recognition tasks. The test result without any engineering has shown the potential for practical application. At the end of the paper we add some related experimental result that has been obtained using a different model - dynamic Bayesian network - which is also a type of stochastic model.

• 한국정밀공학회지
• /
• 제18권11호
• /
• pp.80-85
• /
• 2001

In Part 2, dynamic cutting force model, thermal behavior model, and feed drive model are presented for development of a virtual machine tool. Some relevant results with brief descriptions for each model are presented to verify the proposed models. Experimental results for each model agreed well with the estimated ones. The developed models in this two-part paper are partially integrated as a comprehensive software environment.

• Journal of international Conference on Electrical Machines and Systems
• /
• 제3권4호
• /
• pp.394-397
• /
• 2014

This paper deals with the comparison and dynamic behavior of a moving-coil linear oscillatory actuator (MCLOA) with/without a mechanical spring. On the basis of a dynamic simulation model, the dynamic characteristics such as a current and a stroke of the MCLOA without the spring are predicted for various values of frequency. And then, dynamic test results are given to confirm the dynamic simulations. Finally, this paper describes the influence of the spring on the dynamic behavior of the MCLOA from the dynamic experiments of the MCLOA with/without the spring.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 1998년도 추계학술대회 논문집
• /
• pp.280-287
• /
• 1998

In this paper, Dynamic Fault Tree algorithm(DFT algorithm) is presented. This new algorithm provides a concise representation of dynamic fault tolerance system structure with redundancy, dynamic redundancy management and complex fault ＆ error recovery techniques. And it allows the modeler to define a dynamic fault tree model with the relative advantages of both fault tree and Markov models that captures the system structure and dynamic behavior. This algorithm applies to TMR and Dual-Duplex systems with the dynamic behavior and show that this algorithm captured the dynamic behavior in these systems with fault ＆ error recovery technique, sequence-dependent failures and the use dynamic spare. The DFT algorithm for solving the problems of the systems is more effective than the Markov and Fault tree analysis model.

• 한국생산제조학회지
• /
• 제5권4호
• /
• pp.82-89
• /
• 1996

In this study, to choose the drilling m/c with analysis model for dynamic design of machine tool strctures, are used substucture syntheis method for reduction to degrees of freedom of dynamic model and analysis evaluation of substructures The dynamic factors of substurctures are examined by substructure synthesis method. And that dynamic design of structures for energy balancing are performed. The computer program for calculated of the dynamic and energy distribution analysis was developed. Result of numerical analysis by developed program obtained to conclusion as following. The design of machine tool structures by dynamic avoid the resonances, and are known to considered based on the energy balancing. These methods can be used effectively for the performance evaluation, design modification and improvement of dynamic performance evaluation, design modification and improvement of dynamic performance of machine tools.

• Journal of Electrical Engineering and Technology
• /
• 제1권2호
• /
• pp.161-169
• /
• 2006

Load representation has a significant impact on power system analysis and control results. In this paper, composite load models are developed based on on-line measurement data from a practical power system. Three types of static-dynamic load models are derived: general ZIP-induction motor model, Exponential-induction motor model and Z-induction motor model. For the dynamic induction motor model, two different third-order induction motor models are studied. The performances in modeling real and reactive power behaviors by composite load models are compared with other dynamic load models in terms of relative mismatch error. In addition, numerical consideration of ill-conditioned parameters is addressed based on trajectory sensitivity. Numerical studies indicate that the developed composite load models can accurately capture the dynamic behaviors of loads during disturbance.