• 제어로봇시스템학회논문지
• /
• 제18권3호
• /
• pp.175-180
• /
• 2012

A nonlinear navigation filter for biomimetic robot using analytic approximation of mean and covariance of state variable is proposed. The approximations are performed at the time update step in the filter structure. The mean is approximated to the 3rd order of Taylor's series expansion of true mean and the covariance is approximated to the 3rd order either. The famous EKF is a nonlinear filtering method approximating the mean to 1st order and the covariance to the 3rd order. The UKF approximate them to the higher orders by numerical method. The proposed method derived a analytical approximation of them for navigation system and therefore don't need so called sigma point transformation in UKF. The simulation results show that the proposed method can be a good alternative of UKF in the systems which require less computational burden.

• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 1993년도 Fifth International Fuzzy Systems Association World Congress 93
• /
• pp.865-868
• /
• 1993

This paper describes a simple random signal generator employing by CMOS analog technology in current mode. The system is a nonlinear dynamical system described by a difference equation, such as x(t＋1) = f(x(t)) , t = 0,1,2, ... , where f($.$) is a nonlinear function of x(f). The tent map is used as a nonlinear function to produce the random signals with the uniform distribution. The prototype is implemented by using transistor array devices fabricated in a mass product line. It can be easily realized on a chip. Uniform randomness of the signal is examined by the serial correlation test and the $\chi$2 test.

• Nuclear Engineering and Technology
• /
• 제53권8호
• /
• pp.2696-2707
• /
• 2021

The purpose of this study is to investigate the efficiency of various structural modeling schemes for evaluating seismic performances and fragility of the reactor containment building (RCB) structure in the advanced power reactor 1400 (APR1400) nuclear power plant (NPP). Four structural modeling schemes, i.e. lumped-mass stick model (LMSM), solid-based finite element model (Solid FEM), multi-layer shell model (MLSM), and beam-truss model (BTM), are developed to simulate the seismic behaviors of the containment structure. A full three-dimensional finite element model (full 3D FEM) is additionally constructed to verify the previous numerical models. A set of input ground motions with response spectra matching to the US NRC 1.60 design spectrum is generated to perform linear and nonlinear time-history analyses. Floor response spectra (FRS) and floor displacements are obtained at the different elevations of the structure since they are critical outputs for evaluating the seismic vulnerability of RCB and secondary components. The results show that the difference in seismic responses between linear and nonlinear analyses gets larger as an earthquake intensity increases. It is observed that the linear analysis underestimates floor displacements while it overestimates floor accelerations. Moreover, a systematic assessment of the capability and efficiency of each structural model is presented thoroughly. MLSM can be an alternative approach to a full 3D FEM, which is complicated in modeling and extremely time-consuming in dynamic analyses. Specifically, BTM is recommended as the optimal model for evaluating the nonlinear seismic performance of NPP structures. Thereafter, linear and nonlinear BTM are employed in a series of time-history analyses to develop fragility curves of RCB for different damage states. It is shown that the linear analysis underestimates the probability of damage of RCB at a given earthquake intensity when compared to the nonlinear analysis. The nonlinear analysis approach is highly suggested for assessing the vulnerability of NPP structures.

• 한국수자원학회논문집
• /
• 제31권6호
• /
• pp.769-777
• /
• 1998

본 연구에서는 다양한 특성을 가지는 시계열 자료들을 분석하여 자료의 비선형성 여부를 판단하였다. 시계열 자료의 무작위성을 분석하면 시스템의 비선형 구조를 알아낼 수 있다. 무작위성을 조사하는 통계기법으로는 전통적인 비모수 통계기법과 새로운 통계기법인 BDS 통계를 사용하였으며, 그들의 해석결과를 비교하였다. BDS 통계는 카오스 분석을 위해 이용되는 상관적분의 통계학적 특성을 바탕으로 한 검정방법으로서 무작위성과 비선형 동역학 시스템을 구분하는데 탁월한 능력이 잇는 것으로 알려져 왔다. 이미 자료의 특성이 알려진 선형, 비선형 시스템에 BDS 통계를 적용한 결과, 비모수 통계기법에 비해 더욱 정확한 해석결과를 나타내었다. 실제 수문 시계열 자료를 이용하여 선형 추계학적 모형인 ARMA 형태의 모형을 구축한 후, 이 모형으로부터 계산된 잔치를 BDS 통계를 사용하여 분석하였다. 분석결과, BDS 통계는 시계열자료의 무작위성과 카오스 시스템의 비선형성을 판단하여 줄 뿐만아니라, 추계학적 모형의 잔차 분석을 통한 모형의 적합성 판단에도 유용한 방법임을 알 수 있었다.

• 한국지능시스템학회논문지
• /
• 제17권4호
• /
• pp.563-568
• /
• 2007

본 논문은 가중 퍼지소속함수 기반 신경망(Neural Network with Weighted Fuzzy Membership Functions, NEWFM)을 이용하여 클래스의 분류강도를 구하고 비선형 시계열 추이선을 예측하는 방안을 제안하고 있다. NEWFM에 의하여 추출된 가중퍼지 소속함수(BSWFM)를 이용하여 입력값에 대한 분류강도를 구하게 되고, 이들에 대한 가중평균 역퍼지화를 통하여 비선형 시계열 추이선을 작성한다. 실증분석결과 NEWFM은 목표 클래스로 설정된 GDP에 대하여 92.22%의 분류성능을 보여 주었다. 따라서 동 비선형 시계열 추이선은 대표적인 경기지표인 GDP 추이에 비교적 높은 유사도를 나타내는 가운데 분석대상기간인 제5순환기-제8순환기 중 정점(peak)에서 평균 12개월, 저점(trough)에서 평균 6개월의 선행성(look-ahead)을 보여 줌으로써 경기변동에 앞서 상당기간의 시차를 둔 예측지표로서 활용가능성이 입증되었다. NEWFM은 그 특징선택(feature selection)에 의하여 선행지표 10개 중 3개의 축소를 기할 수 있게 해 줌으로써 보다 적은 수의 경제지표를 가지고도 분류성능을 90.0%에서 92.22%로 향상을 기하는 가운데 효율적인 예측기능을 수행할 수 있음이 입증되었다.

• 대한토목학회논문집
• /
• 제26권3B호
• /
• pp.279-289
• /
• 2006

최근에 수문시계열로부터 저차원의 비선형 거동을 재구성하고자 하는 연구가 활발히 진행되고 있다. 이러한 관점에서 본 연구에서는 Support Vector Machine(SVM)을 이용하여 우수한 상태-공간 재구성 능력을 갖는 비선형 예측모형을 구성하여 Great Salt Lake(GSL) Volume에 적용하였다. SVM은 Kernel 함수로부터 유도된 고차원의 특성공간 안에서 선형함수의 가상공간을 이용하는 Machine Learning 방법론이다. 또한 SVM은 훈련자료로부터 얻어지는 평균제곱오차가 아닌 일반화된 오차를 최소화함으로써 상대적으로 기존 방법에 비해 적은 수의 매개변수와 과적합(over fitting)을 피하면서 비선형 함수의 최적화가 가능하다. 본 연구에서 제시한 SVM 회귀분석의 적용성은 미국의 GSL의 2주 간격 Volume을 대상으로 검토하였다. SVM을 이용한 비선형 예측모형은 GSL Volume의 2주(1-Step), 8주(4-Step)와 반복예측(Iterated Prediction, 121-Step)까지 적용되었다. 본 연구에서는 극치사상 즉, 급격한 감소 및 증가 구간을 예측하는데 있어서 훈련구간과 예측구간을 구분하여 모형의 신뢰성을 평가하였다. 예측결과SVM은 훈련자료로부터 적은 수의 관측치를 이용하여 동역학적 거동을 추출할 수 있었으며 실제 관측자료와 거의 유사한 예측이 가능함을 통계적 지표로 확인할 수 있었다. 따라서 비선형 수문시계열의 단기 예측을 위한 모형으로 적용이 가능할 것으로 판단된다.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제16권3호
• /
• pp.163-172
• /
• 2016

Electric load forecasting is essential for effective power system planning and operation. Complex and nonlinear relationships exist between the electric loads and their exogenous factors. In addition, time-series load data has non-stationary characteristics, such as trend, seasonality and anomalous day effects, making it difficult to predict the future loads. This paper proposes a locally-weighted polynomial neural network (LWPNN), which is a combination of a polynomial neural network (PNN) and locally-weighted regression (LWR) for daily shortterm peak load forecasting. Model over-fitting problems can be prevented effectively because PNN has an automatic structure identification mechanism for nonlinear system modeling. LWR applied to optimize the regression coefficients of LWPNN only uses the locally-weighted learning data points located in the neighborhood of the current query point instead of using all data points. LWPNN is very effective and suitable for predicting an electric load series with nonlinear and non-stationary characteristics. To confirm the effectiveness, the proposed LWPNN, standard PNN, support vector regression and artificial neural network are applied to a real world daily peak load dataset in Korea. The proposed LWPNN shows significantly good prediction accuracy compared to the other methods.

• International Journal of Naval Architecture and Ocean Engineering
• /
• 제7권2호
• /
• pp.301-314
• /
• 2015

Aiming at accurately distinguishing modeless component and natural vibration mode terms from data series of nonlinear and non-stationary processes, such as Vortex-Induced Vibration (VIV), a new empirical mode decomposition method has been developed in this paper. The key innovation related to this technique concerns the method to decompose modeless component from non-stationary process, characterized by a predetermined 'maximum intrinsic time window' and cubic spline. The introduction of conceptual modeless component eliminates the requirement of using spurious harmonics to represent nonlinear and non-stationary signals and then makes subsequent modal identification more accurate and meaningful. It neither slacks the vibration power of natural modes nor aggrandizes spurious energy of modeless component. The scale of the maximum intrinsic time window has been well designed, avoiding energy aliasing in data processing. Finally, it has been applied to analyze data series of vortex-induced vibration processes. Taking advantage of this newly introduced empirical decomposition method and mode identification technique, the vibration analysis about vortex-induced vibration becomes more meaningful.

• 한국운동역학회지
• /
• 제33권1호
• /
• pp.34-44
• /
• 2023

Objective: The purpose of this study was to examine the effects of increasing running speed on human stability by comparing the Lyapunov Exponent (LyE) and Coefficient of Variation (CV) methods, with the goal of identifying key variables and uncovering new insights. Method: Fourteen adult males (age: 24.7 ± 6.4 yrs, height: 176.9 ± 4.6 cm, weight: 74.7 ± 10.9 kg) participated in this study. Results: In the CV method, significant differences were observed in ankle (flexion-inversion/eversion; p < .05) and hip joint (internal-external rotation; p < .05) movements, while the center of mass (COM) variable in the coronal axis movements showed a significant difference at the p < .001 level. In the LyE method, statistical differences were observed at the p < .05 level in knee (flexion-extension), hip joint (internal-external rotation) movements, and COM across all three directions (sagittal, coronal, and transverse axis). Conclusion: Our results revealed that the stability of the human body is affected at faster running speeds. The movement of the COM and ankle joint were identified as the most critical factors influencing stability. This suggests that LyE, a nonlinear time series analysis, should be actively introduced to better understand human stabilization strategies.

• 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.