• 전기학회논문지
• /
• 제61권1호
• /
• pp.162-167
• /
• 2012

In power quality monitoring, it is very important to detect disturbances (sag, swell, transient, and interruption) accurately. In this paper, a detection method for power quality disturbances by combining the filter bank system and adaptive filter is proposed. To decompose power signal, binary tree structured filter bank system is designed. In the filter bank system, the fundamental filter bank(QMF bank) is used as a module in each decomposing level. An adaptive filter is used to improve the detection accuracy of disturbances for each subband signal. In the adaptive filter, the measure of estimated error change is used to detect singular points of power quality disturbances. Computer simulations were performed on synthetic signals which have disturbances to assess the performance of the proposed method.

• 전기학회논문지P
• /
• 제50권3호
• /
• pp.130-137
• /
• 2001

In this paper, we propose a new Active Noise Filter Control System which operate as a control performance when a adaptive filter fault. In this system, half-fixed filter which is new filter, connected to parallel with adaptive filter. An adaptive filler use to continuous parameter estimating, but adaptive filter is fault, half-fixed filter update newly data which is continuous estimating date each during sampling period. We simulate and apply the proposed active noise filter system to in the cylinder type duct. Experimental results show that proposed Active Noise Filter Control System has better control performance than existing filter which Eriksson's or Parallel Filter System in term of noise reduction.

• 한국음향학회지
• /
• 제21권5호
• /
• pp.438-445
• /
• 2002

본 논문에서는 격자 트랜스버설 결합 (LTJ) 적응필터를 시변 변환영역 적응필터 관점에서 해석함으로써 필터계수보상의 필요성을 보다 쉽고 일반적으로 해석하였다. 또한, 음성 신호가 단구간에서 정적이라는 특성을 이용하여 필터계수 보상을 위한 계산량을 감소시킬 수 있는 방법을 제안하였으며, 모의 음성신호와 실제 음성신호를 사용한 실험을 통하여 효용성을 입증하였다. 제안된 적응필터는 필터계수 보상을 위한 계산량이 95% 감소되었으며, 1000탭을 사용하는 음향반향제거기의 경우 전체 시스템의 계산량을 82% 감소시킬 수 있다.

• 전기학회논문지
• /
• 제68권1호
• /
• pp.108-121
• /
• 2019

This paper presents the study results on the design of the correction filter to improve the azimuth error estimation of the high-speed guided missile launched from the warship after the transfer alignment. We theoretically proved that the transfer alignment performance is determined by the accuracy of the marine inertial navigation system and the observability of the attitude error state variable in the transfer alignment filter, and that most of navigation errors in high-speed guided missile are caused by azimuth error. In order to improve the azimuth estimation performance of the correction filter, the multiple adaptive estimation method and the adaptive filters adapting the measurement noise covariance or the process noise covariance are proposed. The azimuth estimation performance of the proposed adaptive filter and the existing Kalman filter are compared and analyzed each other for 8 different transfer alignment accuracy cases. As a result of comparison and analysis, it was confirmed that the adaptive filter adapting the process noise covariance has the best azimuth estimation performance. These results can be applied to the design of correction filters for high-speed guided missile.

• 한국항공우주학회지
• /
• 제36권5호
• /
• pp.455-461
• /
• 2008

본 논문에서는 스칼라 필터와 적응 스칼라 필터의 성능을 비교 분석하였다. 외부 실험 과 비슷한 실험 환경을 만들기 위해서 GPS를 대신할 수 있는 초음파 의사위성을 사용하였다. 적응 스칼라 필터는 스칼라 필터와는 달리 적응기법을 사용하여 연속적으로 속도오차 공분산과 측정잡음 공분산을 추정한다. 실험결과, 적응 기법을 사용하여 위의 두 파라미터를 연속적으로 추정하는 적응 스칼라 필터의 위치 추정 성능이 스칼라 필터보다 더 좋다는 것을 확인하였다.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제55권8호
• /
• pp.375-377
• /
• 2006

An adaptive IIR filter in ANC(Active Noise Control) systems is more effective than an adaptive FIR filter when acoustic feedback exists, in which cause an order of an adaptive FIR filter must be very large if some of poles of the ideal control filter are near the unit circle. But the IIR filters may have stability problems especially when the adaptive algorithm for adaptive filters is not yet converged. In this paper, a stabilized multi-channel recursive LMS (MCRLMS) algorithm for an adaptive multi-channel IIR filter is presented. RLMS algorithms usually diverge before the algorithm is not yet converged. So, in the beginning of the ANC system, the stability of the RLMS algorithms could be improved by pulling the poles of the IIR filter to the center of the unit circle, and returning the poles to their original positions after the filter converges. Computer simulations and experiments for dipole ducts using a TMS320C32 digital signal processor have performed to show the effectiveness of a proposed algorithm.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2006년도 심포지엄 논문집 정보 및 제어부문
• /
• pp.30-32
• /
• 2006

An adaptive IIR filter in ANC(Active Noise Control) systems is more effective than an adaptive FIR filter when acoustic feedback exists, in which cause an order of an adaptive FIR filter must be very large if some of poles of the ideal control filter are near the unit circle. But the IIR filters may have stability problems especially when the adaptive algorithm for adaptive filters is not yet converged. In this paper, a stabilized multi-channel recursive LMS (MCRLMS) algorithm for an adaptive multi-channel IIR filter is presented. RLMS algorithms usually diverge before the algorithm is not yet converged. So, in the beginning of the ANC system, the stability of the RLMS algorithms could be Improved by pulling the poles of the IIR filter to the center of the unit circle, and returning the poles to their original positions after the filter converges. Computer simulations and experiments for dipole ducts using a TMS320C32 digital signal processor have performed to show the effectiveness of a proposed algorithm.

• 한국음향학회지
• /
• 제12권1호
• /
• pp.37-46
• /
• 1993

The adaptive algorithm for the Volterra filter is considered. Owing to its simplicity, the LMS algorithm for adaptive Volterra filter(AVF) is widely used as in linear adaptive filters. However, the convergence speed is unsatisfactory. For improving the convergence speed, the frequency domain LMS second order adaptive Volterra filter(FLMS-AVF) is proposed and analyzed. We show that the time and frequency domain LMS AVF's have the same steady state performance under approprate conditons. Moreover, it can be shown that this algorithm can improve the convergence speed significantly by applying self-orthogonalizing method.

• 제어로봇시스템학회논문지
• /
• 제9권6호
• /
• pp.420-424
• /
• 2003

A new adaptive filter is proposed for an aided strapdown inertial navigation system(SDINS). The proposed filter can be used to effectively estimate the time-varying variance of the measurement noise. Then, the in-flight alignment for the radar aided SDINS is designed using the additive quatermion error model. Simulation results show that the proposed adaptive filter effectively improves the performance of the radar aided SDINS.

• 한국멀티미디어학회논문지
• /
• 제7권12호
• /
• pp.1650-1656
• /
• 2004

In this paper, a multi-adaptive filter is proposed for removing EOG and the 60 Hz power supply noise from EEG measured in the frontal lobe and the feedback output control method is implemented for biofeedback. The multi-adaptive filter has been implemented on the TMS320C6711 DSP system and the feedback output control algorithm has been realized by calculating the ratio of alpha wave on the TMS320C31 DSP system with real time performance. Through the experiment using the implemented multi-adaptive filter and feedback output controller, we demonstrate that the proposed adaptive filter effectively removes EOG and the 60 Hz power supply noise from the measured EEG in the frontal lobe and the feedback algorithm controls the level of stimulation by the ratio of the alpha wave.