• 전자공학회논문지 IE
• /
• 제45권3호
• /
• pp.20-26
• /
• 2008

두 가지 Independent Component Analysis(ICA) 알고리즘을 적용하여 자동차 엔진 소음과 섞인 음성 신호의 인식을 시도하였다. 이를 이용하여 추정한 신호를 HMM을 이용하여 인식하였고 이 신호의 인식률을 소음이 섞이기 전의 음성 신호의 인식률과 비교하였다. 음성 신호를 추정하는데 두 가지 서로 다른 ICA를 사용하였으며 그 중의 하나는 negentropy를 최대화하는 FastICA 알고리즘이며 다른 하나는 출력 신호 사이의 독립성을 최대화하여서 입력과 출력 사이의 mutual information을 최대화하는 information-maximization approach 이다. 남성 앵커가 진행한 한국어 뉴스 문장에 대한 단어 인식률은 87.85%이며 다양한 신호 대 잡음비를 갖도록 소음을 섞어서 추정을 한 후 인식을 시도한 결과 FastICA를 이용해 추정한 음성 신호에 대한 인식률은 1.65%, information-maximization을 이용해 추정한 음성 신호에 대한 인식률은 2.02% 인식률 저하가 나타났다. 따라서 어느 방법을 적용하든지 의미 있는 차이가 없음을 확인하였다.

• 한국정보통신학회논문지
• /
• 제22권12호
• /
• pp.1684-1690
• /
• 2018

근접한 이중대역 신호에 대한 새로운 디지털 전치왜곡 기법을 제안한다. 본 논문에서 고려하는 시스템은 이중대역 신호를 한 개의 전력증폭기로 증폭하는 송신기이다. 이 경우 전력증폭기 출력신호는 두 대역 신호의 교차변조 및 상호변조에 의해 왜곡이 발생한다. 특히 두 대역이 가까울 때에는 각 대역에서 상호변조에 의해 발행한 스펙트럼이 서로 겹칠 수 있고 이는 전치왜곡 성능을 저하시킨다. 이러한 문제를 해결하기 위해서 제안하는 기법은 먼저 전력증폭기 특성을 추정하고 이렇게 추정한 전력증폭기 특성을 바탕으로 전치왜곡 계수를 추출한다. 이러한 두 단계를 통해 전치왜곡 계수를 구하면 근접한 이중대역 신호에 대해서도 서로 간섭 없이 전치왜곡이 동작할 수 있다. 제안하는 기법은 컴퓨터 모의실험을 통해 검증하는데, 모의실험 결과에 따르면 제안하는 기법이 기존의 이중대역 전치왜곡 방법보다 우수한 성능을 보인다.

• 한국음향학회지
• /
• 제40권2호
• /
• pp.161-168
• /
• 2021

본 논문에서는 발성하고 있는 입 주변에 40 kHz의 주파수를 갖는 초음파 신호를 방사하고 되돌아오는 신호의 도플러 변이를 검출하여 발성음을 합성하는 무 음성 대화기술을 제안하였다. 무음성 대화 기술에서는 비 음성 신호로 부터 추출된 특징변수와 해당 음성 신호의 파라메터 간 대응 규칙을 생성하고 이를 이용하여 음성신호를 합성하게 된다. 기존의 무 음성 대화기술에서는 추정된 음성 파라메터와 실제 음성 파라메터간의 오차가 최소화되도록 대응규칙을 생성한다. 본 연구에서는 추정 음성 파라메터가 실제 음성 파라메터의 분포와 유사하도록 생성적 적대 신경망을 도입하여 대응 규칙을 생성하도록 하였다. 60개 한국어 음성을 대상으로 한 실험에서 제안된 기법은 객관적, 주관적 지표상 으로 기존의 신경망 기반 기법보다 우수한 성능을 나타내었다.

• Journal of Electrical Engineering and Technology
• /
• 제12권1호
• /
• pp.378-384
• /
• 2017

In this paper, an efficient short-time Fourier analysis method for the quasi-periodic signals is proposed via an optimal fixed-lag finite impulse response (FIR) smoother approach using a receding horizon scheme. In order to deal with time-varying Fourier coefficients (FCs) of quasi-periodic signals, a state space model including FCs as state variables is augmented with the variants of FCs. Through an optimal fixed-lag FIR smoother, FCs and their increments are estimated simultaneously and combined to produce final estimates. A lag size of the optimal fixed-lag FIR smoother is chosen to minimize the estimation error. Since the proposed estimation scheme carries out the correction process with the estimated variants of FCs, it is highly probable that the smaller estimation error is achieved compared with existing approaches not making use of such a process. It is shown through numerical simulation that the proposed scheme has better tracking ability for estimating time-varying FCs compared with existing ones.

• Journal of Power Electronics
• /
• 제10권6호
• /
• pp.709-716
• /
• 2010

MDPS (motor driven power steering) systems have been widely used in vehicles due to their improved fuel efficiency and steering performance when compared to conventional hydraulic steering. However, the reduction of torque ripples and material cost are important issues. A low resolution position sensor for MDPS is one of the candidates for reducing the material costs. However, it may increases the torque ripple due to the current harmonics caused by low resolution encoder signals. In this paper, the torque ripple caused by the quantized rotor position of the low resolution encoder is analyzed. To reduce the torque ripples caused by the quantization of the encoder signals, the rotor position and the speed are estimated by measuring the frequency of the encoder signals. In addition, the compensating q-axis current is added to the current command so that the 6th order torque harmonic is attenuated. The reduction of torque ripples by applying the estimated rotor position and the compensated q-axis current is verified through experimental results.

• 전기학회논문지
• /
• 제57권9호
• /
• pp.1674-1680
• /
• 2008

The objective of this study is to compare age-related differences of 15 characteristic variables estimated from SEMG signals, which were obtained from the old group(62 subjects) and young group(72 subjects). The SEMG signals were recorded from biceps brachii muscle under sustained isometric elbow flexion contraction with constant load(4kg dumbbell) during 30 seconds. Ten time domain (ARV, RMS, MSA, MTA, ZC, MSF, MTF, MSD, MTD, MSS) and five frequency domain(MNF, MDF, $f_{max}$, $P_{max}$, $P_{to}$) variables were extracted from the SEMG signals and then statistically analyzed, respectively. The findings of the analysis indicate significant age-related differences in the SEMG signal during constant load contraction.

• 대한의용생체공학회:의공학회지
• /
• 제15권4호
• /
• pp.455-464
• /
• 1994

The intelligent trajectory control method that controls moving direction and average velocity for a prosthetic arm is proposed by pattern recognition and force estimations using EMG signals. Also, we propose the real time trajectory planning method which generates continuous accelleration paths using 3 stage linear filters to minimize the impact to human body induced by arm motions and to reduce the muscle fatigue. We use combination of MLP and fuzzy filter for pattern recognition to estimate the direction of a muscle and Hogan's method for the force estimation. EMG signals are acquired by using a amputation simulator and 2 dimensional joystick motion. The simulation results of proposed prosthetic arm control system using the EMG signals show that the arm is effectively followed the desired trajectory depended on estimated force and direction of muscle movements.

• 대한전기학회논문지
• /
• 제45권3호
• /
• pp.432-437
• /
• 1996

This paper describes a diagnostic technique of power transformers by on-line detection of ultrasonic signals. A trend of partial discharge variation in a real transformer was estimated by counting the number of ultrasonic signals until insulation paper in the point-to-plane electrode is punctured. The number of electrical signals is closely related to the ultrasonic signals generated by partial discharge. The trend of the ultrasonic signal number could easily be distinguished by taking moving average. The insulation failure due to partial discharge in transformers can be predicted based on the trend analysis of ultrasonic signal number caused by partial discharge. (author). 14 refs., 10 figs.

• 한국철도학회:학술대회논문집
• /
• 한국철도학회 2003년도 추계학술대회 논문집(III)
• /
• pp.247-251
• /
• 2003

The velocity of train which is measured from pulse generator attached to TM is used for displaying or control signal of inverter and so on. Measured signals increase and decrease step-by-step by pulse counting or monotonously by F/V conversion. But noises and signal distortions by measuring error like alias make it difficult to provide correct velocity infomation and estimate the acceleration. In this paper, we investigated the performance of Smoothing method for suppressing the noises in velocity signals. And the difference between Smoothed signal and origin velocity signals is inspected and the comparison with low pass filtering show applicable of Smoothing method for noise rejection and the estimation of signal. Finally, acceleration curves estimated from Smoothing method are compared with real accelerator signal attached to train.

• Journal of Power Electronics
• /
• 제12권4호
• /
• pp.604-614
• /
• 2012

This paper presents a comparative study of position sensorless control schemes based on back-electromotive force (back-EMF) estimation in permanent magnet synchronous motors (PMSM). The characteristics of the estimated back-EMF signals are analyzed using various mathematical models of a PMSM. The transfer functions of the estimators, based on the extended EMF model in the rotor reference frame, are derived to show their similarity. They are then used for the analysis of the effects of both the motor parameter variations and the voltage errors due to inverter nonlinearity on the accuracy of the back-EMF estimation. The differences between a phase-locked-loop (PLL) type estimator and a Luenberger observer type estimator, generally used for extracting rotor speed and position information from estimated back-EMF signals, are also examined. An experimental study with a 250-W interior-permanent-magnet machine has been performed to validate the analyses.