• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2004년도 학술대회 논문집 정보 및 제어부문
• /
• pp.349-351
• /
• 2004

The classical dynamic backpropagation learning algorithm has the problems of learning speed and the determine of learning parameter. The Extend Kalman Filter(EKF) is used effectively for a state estimation method for a non linear dynamic system. This paper presents a learning algorithm using Dual Extended Kalman Filter(DEKF) for Fully Recurrent Neural Network(FRNN). This DEKF learning algorithm gives the minimum variance estimate of the weights and the hidden outputs. The proposed DEKF learning algorithm is applied to the system identification of a nonlinear SISO system and compared with dynamic backpropagation learning algorithm.

• 응용통계연구
• /
• 제23권6호
• /
• pp.1125-1145
• /
• 2010

For modeling(skewed) semicircular data, we derive a new exponential family of distributions. We extend it to the l-axial exponential family of distributions by a projection for modeling any arc of arbitrary length. It is straightforward to generate samples from the l-axial exponential family of distributions. Asymptotic result reveals that the linear exponential family of distributions can be used to approximate the l-axial exponential family of distributions. Some trigonometric moments are also derived in closed forms. The maximum likelihood estimation is adopted to estimate model parameters. Some hypotheses tests and confidence intervals are also developed. The Kolmogorov-Smirnov test is adopted for a goodness of t test of the l-axial exponential family of distributions. Samples of orientations are used to demonstrate the proposed model.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2011년도 제42회 하계학술대회
• /
• pp.1059-1060
• /
• 2011

This paper deals with optimum design criteria for maximum torque density & minimum torque ripple of Flux Switching Motor (FSM) using RSM & FEM. The focus of this paper is to find a design solution through the comparison of torque density and torque ripple according to rotor shape variations. And then, a central composite design(CCD) mixed resolution is introduced, and analysis of variance (ANOVA) is conducted to determine the significance of the fitted regression model.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2000년도 춘계종합학술대회
• /
• pp.367-370
• /
• 2000

수동형 거리 추정을 위해 다양한 기법들이 논의되고 있다. 바다 속과 같은 신호대 잡음비가 낮은 환경으로 인해 기존의 기법들은 오차발생으로 인한 정확한 음원의 거리 추정이 곤란하다. 그리하여 견인 어레이상에서 부어레이를 구성하여 부어레이의 빔 형성기를 가지고 다중 음원 환경에서 원하는 방향으로 표적과의 거리를 추정한다. 이 때 탐지 방향 이외의 음원 방향의 이득을 최소화시키는 null-constraint 기법을 사용하여 보다 향상된 거리 추정 기법을 제시한다.

• Journal of Positioning, Navigation, and Timing
• /
• 제9권2호
• /
• pp.59-63
• /
• 2020

The antenna arrays are known to be effective for GPS anti-jamming and the performance can be improved further if a dual-polarized antenna array is used. However, when the Minimum Variance Distortionless Response (MVDR) beamformer is used as a signal processing algorithm for the dual-polarized antenna array, the anti-jamming performance can degrade in the presence of errors in the steering vector that is a key factor of the MVDR beamformer. Therefore, in this paper, the effect of the steering vector error on the anti-jamming performance of the dual-polarized antenna array is analyzed by simulations and the result is compared to that of the single-polarized antenna array.

• 대한교통학회지
• /
• 제9권1호
• /
• pp.47-56
• /
• 1991

With respect to stochastic delays at linked signals the solid quantitative information has not been available as yet. On the basis of field data the values of "I" (variance-mean ration of flow) were related with the rate of flow. The stochastic delays with specific "I" values were obtained from the distribution of overflow queue, which were calculated by the use of Markov chains. This examination of the results led to the derivation of a simple method for calculating stochastic dclays through the introduction of "I" into Miller's model. The good agreement was shown between the model and the field. The relationships between the cycle lengths and delays were examinated in a large number of conditions with regard to degree of saturation. signal split and link length. Within the practical range of cycle length uniform delays were dominant and no critical point was found in terms of minimum, delay. In highly saturated conditions however the weight of stochastic delay is noticeable.

• 한국공작기계학회:학술대회논문집
• /
• 한국공작기계학회 2004년도 추계학술대회 논문집
• /
• pp.28-33
• /
• 2004

At present, industry and researchers are looking for ways to reduce the use of lubricants because of ecological and economical reasons. Therefore, metal cutting is to move toward dry cutting or semi-dry cutting. This paper presents an investigation into MQL(Minimum Quantity Lubrication) machining with the objective of deriving the optimum cutting conditions for the turning process of SM45C. To reach these goals several finish turning experiments were carried out, varying cutting speed, feed rate and oil quantity, with MQL. The surface roughness results of tests were measured and the effects of cutting conditions were analyzed by the method of Analysis of Variance(ANOVA). From the experimental results and ANOVA, it is found that a better surface roughness can be obtained by decreasing oil quantity and feed rate.

• 한국통신학회논문지
• /
• 제19권7호
• /
• pp.1407-1413
• /
• 1994

레이다에서 정확한 표적 정보를 얻기 위해서는 우선 클러터나 간섭 신호를 효율적으로 제거하고 표적을 검출하여야 한다. 본 논문에서는 신호를 제한 조건에 의해 구별되는 직교 부공간으로 투사함으로써 최분산 기분을 갖는 최적 검출기를 제한 조건을 갖지 않는 검출기로 변형하였다. 제안된 방법은 기존의 최적검출 알고리듬과 등가이며, 역행렬을 구하지 않고서도 빠른 수렴속도를 얻기위한 Gram-Schmidt 적교화가 가능한 구조를 가지게 된다. 모의 실험을 통해 제안된 검출기가 공간과 주파수 영역에서 적응적으로 잘 동작하고 있음을 보였다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1990년도 하계학술대회 논문집
• /
• pp.363-366
• /
• 1990

In the indirect vector control method, the rotor resistance variation caused by operating temperature change is an undesirable nature. A new adaptation algorithm to compensate for the rotor resistance change based on the on-line estimation of field vector which requires the measurements of stator voltage and rotor speed is presented in this paper. Also minimum variance controller is presented for the adaptive control performance. This algorithm has been tested by simulating the induction machine using a digital computer and the results are discussed.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제53권6호
• /
• pp.446-452
• /
• 2004

Under the condition of a common narrowband target signal and interference signals from several directions, the linearly constrained minimum variance (LCMV) method using the generalized sidelobe canceller (GSC) for adaptive beamforming has been exploited successfully However, in the case of wideband signals, the length of the adaptive filter must be extended. As a result, the complexity of the beamformer increases, which makes real-time implementation difficult. In this paper, we improve the convergence characteristics of the adaptive filter using the transform domain normalized least mean square (NLMS) approach based on the subband GSC structure without the increase of complexity. Besides, the M-MAX algorithm, which is one of various selective coefficient updating methods, is employed in order to remarkably reduce the computational cost without decreasing the convergence quality. With the combination of these methods, we propose a computationally efficient wideband adaptive beamformer and verify its efficiency through a series of simulations.