• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.35C no.6
• /
• pp.67-75
• /
• 1998

Although the nCE(n-th order cross-entropy) error function resolves the incorrect saturation problem of conventional EBP(error back-propagation) algorithm, the performance of MLP's (multilayer perceptrons) trained using the nCE function depends heavily on the order of the nCE function. In this paper, we propose an adaptive learning rate to make the performance of MLP's insensitive to the order of the nCE error. Additionally, we propose a limited error signal of output node to prevent unstable learning due to the adaptive learning rate. The effectiveness of the proposed method is demonstrated in simulations of handwritten digit recognition and thyroid diagnosis tasks.

• Computational Structural Engineering
• /
• v.11 no.1
• /
• pp.261-274
• /
• 1998

In this paper, local and global error estimates for the element-free Galerkin (EFG) method are proposed. The essence of proposed error estimates is to use the difference between the values of the projected stress and these given directly by the EFG solution. The stress projection can be obtained simply by taking product of shape function based on a different domain of influence with the stresses at nodes. In this study, it was found that the effectivity index is optimized if the domain of influence in stress projection procedure is the smallest that retains regularity of the matrices in EFG. Numerical tests are shown for various 1D and 2D examples illustrating the good effectiveness of the proposed error estimator in the global energy norm and in the local error estimates.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.15 no.4
• /
• pp.817-822
• /
• 2011

This paper first proposes the speech recognition algorithm by detection of the speech and noise sections at each frame using a neural network training by back-propagation algorithm, then proposes the spectral subtraction method which removes the noises at each frame according to detection of the speech and noise sections. In this experiment, the performance of the proposed recognition system was evaluated based on the recognition rate using various speeches that are degraded by white noise and car noise. Moreover, experimental results of the noise reduction by the spectral subtraction method demonstrate using the speech and noise sections detecting by the speech recognition algorithm at each frame. Based on measuring signal-to-noise ratio, experiments confirm that the proposed algorithm is effective for the speech by corrupted the noise using signal-to-noise ratio.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.25 no.10
• /
• pp.1263-1273
• /
• 1988

An adaptive nulling technique is presented to synthetically overcome the integrated problems associated with the conventional LMS array in the performances of jammer rejection, convergence rate, misadjustment, and reference signal generation. The proposed method is to remove the target signal from the array input and to eliminate the reference signal prior to minimization processing. The algorithm is constrained to the residue noise level in adaptive processor. Analysis shows effectiveness of the algorithm for coherent and/or incoherent interference rejection, wide dynamic range of convergence factor, rapid adaptation rate, and small mean square error. Simulation results confirm the theoretical prediction.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.9 no.6
• /
• pp.33-42
• /
• 2010

Travel speed is an important parameter in measurement of road traffic. UTIS(Urban Traffic Information System) was developed as a type of section detector. However, UTIS incur errors caused by irregular vehicle trajectories, wireless communication range and so on. This paper suggests a new model that use an error-filtering algorithm to improve the accuracy of travel speed measurements. In the field test, the variance of the percent errors measured by the new model was reduced. Therefore, it can be concluded that the proposed model significantly improves travel speed measuring accuracy.

• Journal of Advanced Navigation Technology
• /
• v.22 no.5
• /
• pp.359-366
• /
• 2018

In this paper, the noise is removed by using CTLS aircraft installed FBG sensor inside the aircraft wing. We suggest a normal wavelet transform scheme with motion - invariant characteristics for noise reduction. In the case of installing FBG sensors inside the composite material as in CTLS, large and small empty spaces and parts or sections are generated between the adhesive layers, and a signal splitting problem occurs. FBG sensor is not affected by noise. but eletromagnetic, light source, light detector and signal processing device are influeced by noise because these are eletronic components what affected by eletromagnetic wave. because of this, errors are occured. Experimental results show that the noise can be removed using normal wavelet transform and more accurate data detection is possible.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.1
• /
• pp.42-46
• /
• 2002

In this paper, a new pre-processing algorithm is proposed to improve the performance of stereo acoustic echo canceller. The proposed algorithm has the improved performance by the estimation error reduction of filter coefficient using input signal which was reduced reverberation of room in the basis MINT (Mu1tip1e-input/output Inverse Theorem) filtering. For real stereo speech signal and real room impulse response the results of simulation, we showed that the proposed method could improved 3∼5 dB ERLE (Echo Return Loss Enhancement) regardless of NLMS (Normalized Least Mean Square) and Projection adaptive algorithm.

• Proceedings of the KSRS Conference
• /
• 2006.03a
• /
• pp.113-116
• /
• 2006

지표 변수는 지면 근처의 기후변화 및 상태를 파악하는데 중요한 역할을 하기 때문에, 충분히 높은 정확성을 가진 값이 산출되어야 한다. 하지만 이러한 지표 변수는 구름과 눈, 그리고 강수등에 의해서 그 값이 변화하게 된다. 이러한 오차 값을 줄이기 위해 구름제거, 지리보정, 대기보정 등의 위성 전처리 과정이 수행되었다. 하지만 위성 전처리 과정을 수행한 이후에도 정규식생지수 시계열 자료에는 여전히 노이즈가 남아 있기 때문에 이전에 연구에서는 이동 평균등과 같은 다양한 방법으로 노이즈를 제거하고자 하였다. 하지만 이동평균 방법은 참값에 가까운 최고값도 제거하기 때문에 문제점을 가지고 있다. 본 연구에서는 다중 다항회귀식을 이용하여 정규식생지수 시계열 자료 산출시 발생하는 노이즈를 제거 하였다.

• Journal of the Korean Institute of Telematics and Electronics D
• /
• v.35D no.10
• /
• pp.13-21
• /
• 1998

In this paper, we analyze the main-carrier cancellation characteristics of the nulling loop control circuit which is used for the main-carrier cancellation circuit of the feedforward linear power amplifier. A new nulling loop error control method is proposed to improve the linear power amplifier characteristics. With this analysis, the main carrier cancellation ratio can be estimated and the required specifications of the main and auxiliary amplifiers can be optimized for the economic and power efficiency.

• Proceedings of the KIPE Conference
• /
• 2016.07a
• /
• pp.101-102
• /
• 2016

반도체 변압기를 구성하는 단상 계통 연계형 인버터는 역률을 높이기 위해 전류를 계통 전압과 동상으로 제어하는 것이 요구된다. 산업에서 통상적으로 이용되는 비례-적분 제어기는 정현파의 참조 값을 추적할 때 항상 정상 상태 오차를 수반한다. 비례-공진 제어기는 기본 주파수에서 충분히 큰 이득을 도입함으로써 정상 상태 오차를 줄일 수 있다. 그러나 계단 응답에서 정상 상태 오차를 가진다는 단점이 있다. 본 논문에서는 단상 계통 연계형 인버터의 효과적인 전류 제어를 위하여 정현파 추적뿐만 아니라 계단 응답에서 정상 상태 오차 제거가 가능한 제어기를 제안한다. 본 논문에서 제안하는 제어기는 시뮬레이션을 통해 유효성을 검증하였다.