• 대한음성학회지:말소리
• /
• 제55권
• /
• pp.131-142
• /
• 2005

A mismatch between the training and the test conditions often causes a drastic decrease in the performance of the speech recognition systems. In this paper, non-linear transformation techniques based on histogram equalization in the acoustic feature space are studied for reducing the mismatched condition. The purpose of histogram equalization(HEQ) is to convert the probability distribution of test speech into the probability distribution of training speech. While conventional histogram equalization methods consider only the probability distribution of a test speech, for noise-corrupted test speech, its probability distribution is also distorted. The transformation function obtained by this distorted probability distribution maybe bring about miss-transformation of feature vectors, and this causes the performance of histogram equalization to decrease. Therefore, this paper proposes a new method of calculating noise-removed probability distribution by using assumption that the CDF of noisy speech feature vectors consists of component of speech feature vectors and component of noise feature vectors, and this compensated probability distribution is used in HEQ process. In the AURORA-2 framework, the proposed method reduced the error rate by over $44\%$ in clean training condition compared to the baseline system. For multi training condition, the proposed methods are also better than the baseline system.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2006년도 하계종합학술대회
• /
• pp.123-124
• /
• 2006

This paper investigates an efficient pre-equalization technique that can be applied on the up-link of a time division duplex (TDD) orthogonal frequency division multiplexing (OFDM) systems. This technique provides the spectral efficiency by omitting the usage of training sequences on the up-link. In this paper, the performance of a variety of pre-equalization techniques is analyzed and simulations are performed by applying power efficient pre-equalization algorithms with phase and amplitude & phase pre-equalization in the Rayleigh fading channel.

• 한국산학기술학회:학술대회논문집
• /
• 한국산학기술학회 2007년도 추계학술발표논문집
• /
• pp.241-244
• /
• 2007

본 논문에서는 B-FWA의 상향링크에 적용할 수 있는 equalization 시스템 중에서 pre-equalization 시스템에 대해서 다양한 감쇄 채널에 대해서 성능을 분석하였다. 고려한 감쇄채널은 세가지의 강우감쇄 채널에 (rain, intermittent light rain and thundershower) 대해서 분석하였다. 본 논문에서 제안하는 pre-equalization 시스템에 적용된 adaptive 알고리즘은 LMS와 RLS로서, 각각의 트레이닝 시퀀스의 길이 대비 채널환경에 따른 성능을 평가하였다. 실험결과, LMS 알고리즘을 사용하는 equalization 알고리즘은 알고리즘의 복잡도를 어느 정도 허용되는 범위에서 delay에 심한 부분에 적합하는 것으로 결론지었고, RLS 알고리즘은 짧은 트레이닝 시퀀스로 빠른 송신이 가능한 곳에 성능이 우수함을 확인하였다.

• 전력전자학회:학술대회논문집
• /
• 전력전자학회 2018년도 추계학술대회
• /
• pp.36-38
• /
• 2018

As the battery capacity requirement increases, battery cells are connected in a parallel configuration. However, the sharing current of each battery cell becomes unequal due to the imbalance between cell's impedance which results the mismatched states of charge (SOC). The conventional fixed-resistance balancing methods have a limitation in battery equalization performance and system efficiency. This paper proposes a battery equalization method based on dynamic resistance technique, which can improve equalization performance and reduce the loss dissipation. Based on the SOC rate of parallel connected battery cells, the switches in the equalization circuit are controlled to change the equivalent series impedance of the parallel branch, which regulates the current flow to maximize SOC utilization. To verify the method, operations of 4 parallel-connected 18650 Li-ion battery cells with 3.7V-2.6Ah individually are simulated on Matlab/Simulink. The results show that the SOCs are balanced within 1% difference with less power dissipation over the conventional method.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제53권2호
• /
• pp.76-85
• /
• 2004

In this paper, the contrast enhancement method of thermal image is proposed and it is the plateau equalization algorithm using local histogram for the real time display of infrared imagery. Through hardware implementing, its practicality and adequacy are proved. Examinations are executed to verify the effect of contrast enhancement by bright control and contrast control automatic to the plateau value in the manual mode, and that verified the effect of contrast enhancement in the automatic mode and the practicality in the real system. According to the experiment results, the proposed "the application of local histogram and plateau equalization algorithm for contrast enhancement of real time thermal image"in this dissertation is the verified method for the thermal imaging contrast enhancement.

• 전력전자학회:학술대회논문집
• /
• 전력전자학회 2007년도 하계학술대회 논문집
• /
• pp.347-349
• /
• 2007

Charge equalization scheme for HEV lithium-ion battery system is proposed in this paper, where all the primary windings with in parallel bi-directional switches are coupled in series to provide the equalizing energy from the whole battery string to the specific under charged cells. Moreover, to realize minimized size of equalization circuit employing the proposed cell balancing scheme, the optimal power rating design rule according to equalization time and SOC distribution of imbalance is proposed. A prototype of HEV lithium-ion battery system of four cells shows the outstanding charge equalization performance while maintaining greatly reduced size of cell balancing circuit.

• 한국멀티미디어학회논문지
• /
• 제12권9호
• /
• pp.1316-1322
• /
• 2009

광대역 통신을 위한 효율적 전송 방식으로 다중 반송파 시스템이 각광을 받고 있다. Filtered multitone(FMT) 시스템은 orthogonal frequency division multplexing(OFDM)과 달리 시간영역에서 필터의 길이가 여러 심볼 주기에 걸쳐 있으며, 이러한 긴 주기의 필터 사용으로 다중경로 환경에서 intersymbol interference(ISI)에 의한 성능열화 현상이 발생한다. 본 논문에서는 무선 채널 환경에서 필터에 의하여 발생한 ISI 성분을 제거하는 FMT 시스템을 위한 간단한 등화기법을 제안한다. 제안하는 등화기법은 직전 심볼과 직후 심볼에서 발생하는 ISI값을 추정하여, 현재 심볼에서 추정된 값을 빼줌으로써 ISI 성분을 제거해주는 방식으로 동작한다. 제안된 등화기의 성능을 이론적으로 분석하여 성능의 우수함을 보였으며, 컴퓨터 시뮬레이션을 통하여 제안된 시스템의 성능이 lower bound에 근접함을 보여주었다.

• 산업경영시스템학회지
• /
• 제31권4호
• /
• pp.140-145
• /
• 2008

Fuzzy partition is a conceptual vehicle that encapsulates data into information granules. Fuzzy equalization concerns a process of building information granules that are semantically and experimentally meaningful. A few algorithms generating fuzzy partitions with fuzzy equalization have been suggested. Simulations and experiments have showed that fuzzy partition representing more characteristics of given input distribution usually produces meaningful results. In this paper, given two points and cardinality of fuzzy partition, we prove that it is not true that there always exists a fuzzy partition with fuzzy equalization in which two of points having peaks fall on the given two points. Then, we establish an algorithm that minimizes the maximum distance between given two points and adjacent points having peaks in the partition. A numerical example is presented to show the validity of the suggested algorithm.

• 한국컴퓨터정보학회논문지
• /
• 제24권2호
• /
• pp.57-66
• /
• 2019

This paper compared the performance with an overview of channel equalization techniques used in high-speed serial transceivers, including the homogeneous architecture and associated components for the GHz interconnect of backplane and cable channels. It also used the MATLAB tool to present system analysis and simulation results for continuous time equivalent structures. In the case of conventional continuous equalization, high frequency deficits occur due to the use of a comparator that is difficult to implement as well as the low speed limit. In this paper, the channel equalization technique based on the power spectrum analysis of clocks was used to compensate for the frequency loss, and the application of the TX+Channel and TX+Equalizer filters enabled the measurement of attenuation and equivalence without comparators. The application of blender and band-pass filters at high speeds also showed significant effectiveness.

• 한국지능시스템학회논문지
• /
• 제11권9호
• /
• pp.827-832
• /
• 2001

본 논문은 수치적인 데이터를 이용하여 시스템을 구성하는 퍼지 모델링에서 각각의 장점들을 유지하면서 순차적으로 성능을 개선하는 방법을 제안한다. 기존의 다양한 퍼지 모델링의 최적화 방법들은 각각의 뛰어난 최적화 기법을 이용하면서도 순차적으로 퍼지 모델의 성능을 개선하려하는 시도는 많지 않았다. 이에 본 논문에서는 각 단계별로 최적의 성능을 구현하고 이를 다음 단계에서 초기로 이용함으로써 퍼지 모델의 성능이 순차적으로 개선되는 것을 제안하였다. 이는 각각의 최적화 기법들을 지속적으로 이용함으로써 원하는 모델의 성능을 개선하고자 하는 것이다. 제안된 방법의 유용성을 Rice taste 데이터 모델에 적용하여 제안된 방법이 이전의 연구보다 좋은 결과를 보임을 알았다.