• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2005년도 ICCAS
• /
• pp.920-924
• /
• 2005

This paper presents adaptive learning data of evolvable neural networks (ENNs) for time series prediction of nonlinear dynamic systems. ENNs are a special class of neural networks that adopt the concept of biological evolution as a mechanism of adaptation or learning. ENNs can adapt to an environment as well as changes in the environment. ENNs used in this paper are L-system and DNA coding based ENNs. The ENNs adopt the evolution of simultaneous network architecture and weights using indirect encoding. In general just previous data are used for training the predictor that predicts future data. However the characteristics of data and appropriate size of learning data are usually unknown. Therefore we propose adaptive change of learning data size to predict the future data effectively. In order to verify the effectiveness of our scheme, we apply it to chaotic time series predictions of Mackey-Glass data.

• 한국지능시스템학회논문지
• /
• 제16권4호
• /
• pp.396-401
• /
• 2006

순환결합형 신경회로망은 복수 개의 리미트사이클을 생성하며 따라서, 많은 동적 정보를 저장할 수 있는 메모리 시스템으로 사용할 수 있다는 것이 알려져 있다. 본 논문에서는 각 뉴런이 최근접 뉴런에만 이진화한 결합하중 ${\pm}1$로 연결된 연속 시간모델 순환결합형 신경회로망을 구현하였다. 그리고 이런 회로망을 통해 생성되는 리미트사이클의 수와 패턴을 시뮬레이션을 통하여 나타내었다. 또한 카오스 신호를 인가하여 리미트사이클 사이의 천이 가능성을 입증하였다. 특히, 카오스 신호 이외의 랜덤 노이즈를 이용한 해석을 통하여 동적 신경회로망에 카오스 노이즈를 인가하는 경우의 유효성을 검토하였다.

• International Journal of Control, Automation, and Systems
• /
• 제6권5호
• /
• pp.755-766
• /
• 2008

This study presents a recurrent compensatory fuzzy neural network (RCFNN) for dynamic system identification. The proposed RCFNN uses a compensatory fuzzy reasoning method, and has feedback connections added to the rule layer of the RCFNN. The compensatory fuzzy reasoning method can make the fuzzy logic system more effective, and the additional feedback connections can solve temporal problems as well. Moreover, an online learning algorithm is demonstrated to automatically construct the RCFNN. The RCFNN initially contains no rules. The rules are created and adapted as online learning proceeds via simultaneous structure and parameter learning. Structure learning is based on the measure of degree and parameter learning is based on the gradient descent algorithm. The simulation results from identifying dynamic systems demonstrate that the convergence speed of the proposed method exceeds that of conventional methods. Moreover, the number of adjustable parameters of the proposed method is less than the other recurrent methods.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제53권8호
• /
• pp.551-560
• /
• 2004

In this paper, we propose a new architecture of Self-Organizing Fuzzy Polynomial Neural Networks (SOFPNN) that is based on a genetically optimized multilayer perceptron with fuzzy polynomial neurons (FPNs) and discuss its comprehensive design methodology involving mechanisms of genetic optimization, especially genetic algorithms (GAs). The proposed SOFPNN gives rise to a structurally optimized structure and comes with a substantial level of flexibility in comparison to the one we encounter in conventional SOFPNNs. The design procedure applied in the construction of each layer of a SOFPNN deals with its structural optimization involving the selection of preferred nodes (or FPNs) with specific local characteristics (such as the number of input variables, the order of the polynomial of the consequent part of fuzzy rules, and a collection of the specific subset of input variables) and addresses specific aspects of parametric optimization. Through the consecutive process of such structural and parametric optimization, an optimized and flexible fuzzy neural network is generated in a dynamic fashion. To evaluate the performance of the genetically optimized SOFPNN, the model is experimented with using two time series data(gas furnace and chaotic time series), A comparative analysis reveals that the proposed SOFPNN exhibits higher accuracy and superb predictive capability in comparison to some previous models available in the literatures.

• 전자공학회논문지S
• /
• 제35S권10호
• /
• pp.166-174
• /
• 1998

신경망은 분산된 비선형 처리구조와 학습능력 때문에 높은 차수의 비선형 동특성 구현능력을 갖고 있으므로 비정적 신호에 대한 적응예측을 수행할 수 있다. 본 논문에서는 두 가지 방법 (비선형 모듈구조와 비선형과 선형모듈이 직렬로 연결된 예측구조)으로 비정적 신호의 비선형 예측을 다루고 있다. 완전 궤환된 리커런트 신경망과 기존의 TDL(tapped-delay-line) 필터가 비선형과 선형모듈로 각각 사용되었다. 제안된 예측기의 동특성은 카오스 시계열과 음성신호에 대해 시험하였으며, 예측성능의 상대적인 비교를 위해 기존의 ARMA(autoregressive moving average) 구조의 선형 예측모델과 비교하였다. 실험결과에 의하면 신경망을 이용한 적응 예측기는 선형 예측기보다 예측성능이 훨씬 우수하였으며, 특히 직렬구조의 예측기는 신호가 크게 변화하는 시계열의 예측에 효과적으로 사용할 수 있음을 확인하였다.

• 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.

• 한국해양학회지:바다
• /
• 제28권4호
• /
• pp.133-142
• /
• 2023

이 논문에서는 RNN (Recurrent Neural Networks)-LSTM (Long Short-Term Memory) 을 적용하여 Lorenz 시스템을 예측하는 자료 기반 인공지능 모델을 구축하고, 이 모델이 미분방정식을 차분화하여 해를 구하는 역학 모델을 대체할 수 있는지 가능성을 진단하였다. 구축된 자료기반 모델이 초기 조건의 작은 교란이 근본적으로 다른 결과를 만들어내는 Lorenz 시스템의 카오스적인 특성을 반영한다는 것과, 시스템의 안정적인 두 개의 닻을 중심으로 운동하면서 전이 과정을 반복하는 특성, "결정론적 불규칙 흐름"의 특성, 분기 현상을 모사한다는 것을 확인하였다. 또한, 적분 시간 간격을 조절함으로써 전산자원을 절감할 수 있는 자료기반 모델의 장점을 보였다. 향후 자료기반 모델의 정교화와 자료기반 모델을 위한 자료동화 기법의 연구를 통해 자료기반 인공지능 모델의 활용성을 확대할 수 있을 것으로 기대한다.