• 지능정보연구
• /
• 제5권1호
• /
• pp.103-123
• /
• 1999

Input filtering as a preprocessing method is so much crucial to get good performance in time series forecasting. There are a few preprocessing methods (i.e. ARMA outputs as time domain filters, and Fourier transform or wavelet transform as time-frequency domain filters) for handling time series. Specially, the time-frequency domain filters describe the fractal structure of financial markets better than the time domain filters due to theoretically additional frequency information. Therefore, we, first of all, try to describe and analyze specially some issues on the effectiveness of different filtering methods from viewpoint of the performance of a neural network based forecasting. And then we discuss about neural network model architecture issues, for example, what type of neural network learning architecture is selected for our time series forecasting, and what input size should be applied to a model. In this study an input selection problem is limited to a size selection of the lagged input variables. To solve this problem, we simulate on analyzing and comparing a few neural networks having different model architecture and also use an embedding dimension measure as chaotic time series analysis or nonlinear dynamic analysis to reduce the dimensionality (i.e. the size of time delayed input variables) of the models. Throughout our study, experiments for integration methods of joint time-frequency analysis and neural network techniques are applied to a case study of daily Korean won / U. S dollar exchange returns and finally we suggest an integration framework for future research from our experimental results.

• 전기학회논문지
• /
• 제63권7호
• /
• pp.1001-1007
• /
• 2014

This paper proposes a method to reduce operation time using discrete cosine transform and to improve image quality by the DC gain correction. Conventional filtered back projection (FBP) filtering in the frequency domain using Fourier transform, but the filtering process uses complex number operations. To simplify the filtering process, we propose a filtering process using discrete cosine transform. In addition, the image quality of reconstructed images are improved by correcting DC gain of sinograms. To correct the DC gain, we propose to find an optimum DC weight is defined as the ratio of sinogram DC and optimum DC. Experimental results show that the proposed method gets better performance than the conventional method for phantom and clinical CT images.

• 한국공간정보시스템학회:학술대회논문집
• /
• 한국공간정보시스템학회 2004년도 춘계 학술발표회 논문집 제1권1호(통권1호)
• /
• pp.29-36
• /
• 2004

In this paper, a study of damage detection for 2-Dimensional Truss Structures using the parametric projection filter theory is presented. Many researchers are interested in inverse problem and one of solution procedures for inverse problems that are very effective is the approach using the filtering algorithm in conjunction with numerical solution methods. In filtering algorithm, the Kalman filtering algorithm is well known and have been applied to many kind of inverse problems. In this paper, the Parametric projection filtering in conjunction with structural analysis is applied to the identification of damages in 2-D truss structures. The natural frequency and modes of damaged truss model are adopted as the measurement data. The effectiveness of proposed method is verified through the numerical examples.

• 음성과학
• /
• 제1권
• /
• pp.277-283
• /
• 1997

In this paper, we present a robust recognition algorithm based on the Wiener filtering method as a research tool to develop the Korean Speech recognition system. We especially used Wiener filtering method in cepstrum-domain, because the method in frequency-domain is computationally expensive and complex. Evaluation of the effectiveness of this method has been conducted in speaker-independent isolated Korean digit recognition tasks using discrete HMM speech recognition systems. In these tasks, we used 12th order weighted cepstral as a feature vector and added computer simulated white gaussian noise of different levels to clean speech signals for recognition experiments under noisy conditions. Experimental results show that the presented algorithm can provide an improvement in recognition of as much as from $5\%\;to\;\20\%$ in comparison to spectral subtraction method.

• Journal of Communications and Networks
• /
• 제18권1호
• /
• pp.8-18
• /
• 2016

In this paper, we propose a number of blind equalization approaches for time-varying andmulti-path channels. The approaches employ cost reference particle filter (CRPF) as the symbol estimator, and additionally employ either least mean squares algorithm, recursive least squares algorithm, or $H{\infty}$ filter (HF) as a channel estimator such that they are jointly employed for the strategy of "Rao-Blackwellization," or equally called "mixture filtering." The novel feature of the proposed approaches is that the blind equalization is performed based on direct channel estimation with unknown noise statistics of the received signals and channel state system while the channel is not directly estimated in the conventional method, and the noise information if known in similar Kalman mixture filtering approach. Simulation results show that the proposed approaches estimate the transmitted symbols and time-varying channel very effectively, and outperform the previously proposed approach which requires the noise information in its application.

• 한국음향학회지
• /
• 제34권4호
• /
• pp.316-320
• /
• 2015

본 논문에서는 Cepstral Mean Normalization(CMN)과 Cepstral Mean and Variance Normalization(CMVN) 프레임워크에서 극점 필터링(pole filtering) 개념을 Mel-Frequency Cepstral Coefficient(MFCC) 특징 벡터에 적용한다. 또한 분산 정규화를 대신하여 스케일 정규화를 사용하는 Cepstral Mean and Scale Normalization(CMSN)의 성능을 잡음 환경 음성인식 실험을 통해 평가한다. CMN과 CMVN은 보통 발화 단위로 수행되기 때문에 짧은 발화의 경우 특징에 대한 평균과 분산의 추정 신뢰도가 보장되지 않는 문제점을 가지는데, 극점 필터링과 스케일 정규화 방식을 적용함으로 이러한 문제점을 보완할 수 있다. Aurora 2 데이터베이스를 이용한 실험 결과, 극점 필터링과 스케일 정규화를 결합한 특징 정규화 방식의 성능이 가장 높은 성능 향상을 보인다.

• 제어로봇시스템학회논문지
• /
• 제6권12호
• /
• pp.1113-1119
• /
• 2000

To operate a process plant safely and economically, process monitoring is very important. Process monitoring is the task to identify the state of the system from sensor data. Process monitoring includes data acquisition, regulatory control, data reconciliation, fault detection, etc. This research focuses on the data recon-ciliation using scale-space filtering and fault detection using functional-link associative neural networks. Scale-space filtering is a multi-resolution signal analysis method. Scale-space filtering can extract highest frequency factors(noise) effectively. But scale-space filtering has too large calculation costs and end effect problems. This research reduces the calculation cost of scale-space filtering by applying the minimum limit to the gaussian kernel. And the end-effect that occurs at the end of the signal of the scale-space filtering is overcome by using extrapolation related with the clustering change detection method. Nonlinear principal component analysis methods using neural network have been reviewed and the separately expanded functional-link associative neural network is proposed for chemical process monitoring. The separately expanded functional-link associative neural network has better learning capabilities, generalization abilities and short learning time than the exiting-neural networks. Separately expanded functional-link associative neural network can express a statistical model similar to real process by expanding the input data separately. Combining the proposed methods-modified scale-space filtering and fault detection method using the separately expanded functional-link associative neural network-a process monitoring system is proposed in this research. the usefulness of the proposed method is proven by its application a boiler water supply unit.

• 한국해양공학회지
• /
• 제2권2호
• /
• pp.37-45
• /
• 1988

A dynamically positioned vessel must be capable of maintaining a specified position and direction by controlling the thruster devices. The motions of a vessel are often assuned to tne sum of low frequency(LF)motions and high frequency(HF)motions. The former is mainly due to wind, current and second order wave forces, while the latter is mainly due to first order wave forces. In order to avoid the high frequency thruser modulation, the control system must include filters to estimate the low frequency motions from the measured motion signals, This paper presents a control system based on Kalman filtering technique and optimal control tyeory. Using the combined kalmam filter, LF motion estimates and HF ones are achieved from the motion measurement of the vessel. The estimated low frequency motions are used as inputs to the dynamic positioning system. The thruster modulation is minimized using the optimal control theory; Linear Quadratic Gaussian(LQG)controller. The performances of the Kalman filter and the dynamic positioned vessel are investigated by computer simulation.

• 한국콘텐츠학회논문지
• /
• 제8권1호
• /
• pp.282-289
• /
• 2008

협력적 필터링을 개선하기 위하여 많은 기술들이 개발되고 실용화되었으나 아이템의 연관 관계를 정확하게 반영하지는 못한다. 본 논문에서는 협력적 필터링의 문제점을 보완하기 위하여 단어 빈도와 ${\alpha}$-cut에 의한 연관 웹문서 분류를 이용한 추천 시스템을 제안한다. 제안된 방법은 형태소 분석을 통한 웹문서에서 단어를 추출하고 빈도 가중치를 계산한다. 추출된 단어를 Apriori 알고리즘을 이용해서 연관 규칙을 생성하고 신뢰도에 단어 빈도 가중치를 적용한다. 그리고 연관 규칙 하이퍼그래프 분할을 이용하여 연관 단어간의 유사도를 계산한다. 마지막으로 유사 클래스를 기반으로 연관 웹문서를 ${\alpha}$-cut을 이용하여 분류하고 개선된 코사인 유사도를 이용하여 유사도를 계산한다. 실험 결과 제안한 방법이 기존의 방법들보다 우수함을 확인하였다.

• Journal of Biosystems Engineering
• /
• 제14권4호
• /
• pp.229-241
• /
• 1989

The purposes of this study were to develop an electronic-hydraulic draft control system for tractor implements, to investigate the control performance of the system and the possibility of adaptation to the conventional tractor. Experiments were carried out to investigate the responses of the system to the step and sinusoidal inputs in draft control. The effects of control mode, hydraulic flow rate, reference deadband, and proportional constant on control performance of the system were investigated. Moreover, the effects of filtering signals from draft sensor were also investigated. The following conclusions were derived from the study; 1. In draft control, there were hunting problems in controlling the implement without filtering the draft signals. Filtering was performed by a control program of electronic controller and the control performance and stability of the system were improved significantly. 2. For the draft control system operated on on-off control mode, draft was controlled within ${\pm}27-{\pm}55kg_f$ to the reference draft when the hydraulic flow rates were 5-15 l/min. For the draft control system operated on PWM control, draft was controlled within ${\pm}27kg_f$ to the reference draft regardless of hydraulic flow rates. 3. In the frequency responses of the draft control system, control performance on PWM control mode was not better than on on-off control mode because of characteristics of hydraulic valve and drafe sensor. As the hydraulic flow rates increased for the system operated on on-off control mode, the corner frequency of amplitude attenuation increased, but the corner frequency of phase-angle change remained nearly the same. But, the system was unstable beyond the frequency of 3.1 rad/s. 4. The electronic-hydraulic hitch control system developed in this study showed superior control performance, stability and convenience compared to conventional mechanical-hydraulic hitch control system. It is considered to be a superior replacement for the conventional mechanical-hydraulic hitch control system.