• Proceedings of the IEEK Conference
• /
• 2009.05a
• /
• pp.261-263
• /
• 2009

본 연구에서는 복잡한 비선형 모델링 방법인 RBF 뉴럴 네트워크(Radial Basis Function Neural Network)와 PNN(Polynomial Neural Network)을 접목한 새로운 형태의 Radial Basis Function Polynomial Neural Network(RPNN)를 제안한다. RBF 뉴럴 네트워크는 빠른 학습 시간, 일반화 그리고 단순화의 특징으로 비선형 시스템 모델링 등에 적용되고 있으며, PNN은 생성된 노드들 중에서 우수한 결과값을 가진 노드들을 선택함으로써 모델의 근사화 및 일반화에 탁월한 효과를 가진 비선형 모델링 방법이다. 제안된 RPNN모델의 기본적인 구조는 PNN의 형태를 이루고 있으며, 각각의 노드는 RBF 뉴럴 네트워크로 구성하였다. 사용된 RBF 뉴럴 네트워크에서의 커널 함수로는 FCM 클러스터링을 사용하였으며, 각 노드의 후반부는 다항식 구조로 표현하였다. 또한 각 노드의 후반부 파라미터들은 최소자승법을 이용하여 최적화 하였다. 제안한 모델의 적용 및 유용성을 비교 평가하기 위하여 비선형 데이터를 이용하여 그 우수성을 보인다.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.12 no.2
• /
• pp.141-148
• /
• 1999

변위 근거 유한요소해석을 사용하는 대부분의 위상 최적화 기법은 요소의 안정성 부족으로 인하여 체스판 무늬가 주기적 형태로 반복하여 설계영역 내부에 나타난다. 본 연구에서는 선형요소를 이용하면서 최적화 알고리즘의 안정성에 영향을 주지 않고 간단하게 모든 최적화 알고리즘에 이용 가능한 체스판무늬 형성 방지책을 개발하였다. 본 연구의 체스판무늬 형성 방치책에서는 먼저 각 선형요소를 구성하는 절점들의 부치분율을 설계변수로 선정하고, 요소내부의 부피분율을 설계변수로 표현하기 위한 선형 보간함수로 선형요소들의 형상함수를 선정하였다. 그리고, 설계변수와 등가 재료상수와의 상관 관계식은 평균장 근사이론을 이용하여 균질화된 재료에 벌칙인자가 도입된 관계식을 이용하였다. 또한, 본 연구에서는 순차이차계획법인 PLBA 알고리즘을 이용하여 위상 최적화문제를 해석하였다.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.12
• /
• pp.20-26
• /
• 1999

In this paper, a multiple target angle tracking algorithm that can avoid data association problem and has a simple structure is proposed by obtaining the angular innovation of the targets from a signal subspace. The signal subspace is recursively estimated by a signal subspace tracking algorithm, such as PAST. A nonlinear matrix equation which satisfy the estimated signal subspace and the angular innovation is induced and expanded into a Taylor series for linear approximation. The angular innovation is obtained by solving the approximated linear matrix equation in the least square sense. The good performance of the proposed algorithm is demonstrated by various computer simulations.

• Journal of the Institute of Convergence Signal Processing
• /
• v.2 no.2
• /
• pp.95-102
• /
• 2001

Fuzzy system is capable of uniformly approximating any nonlinear function over compact input space. The applications of fuzzy system, however, have been primarily limited by the need for large number of fuzzy rules, in particular, for the high-order nonlinear system. In this paper, we propose the reconstruction methods of fuzzy systems, parallel type and cascade, based on the decomposition of some classes of high-order nonlinear functions. Using the both types appropriately, we can reduce the number of fuzzy rules geometrically. It can be applied to the fuzzy system that has an online adaptive structure. Two examples of adaptive fuzzy sliding mode control are shown in the computer simulations to verify the validity of the proposed algorithm.

• Journal of the Earthquake Engineering Society of Korea
• /
• v.6 no.3
• /
• pp.53-62
• /
• 2002

Site soil condition affects significantly on the seismic response of a structure and is a critical factor for the performance based seismic design of a structure. In this paper, the effects of nonlinear soil properties on the elastic response spectra of a structure including the nonlinearity of a soil due to the earthquake excitation is investigated using one step finite element approach for the entire soil structure system and approximate linear iterative procedure to simulate the nonlinear soil behavior with the Ramberg-Osgood soil model. Studies were carried out for a linear SDOF system of a variable period with and without a pile group for the 1940 CI Centro earthquake recorded on ground rather than rock. The study results showed clearly that the effect of the nonlinear behavior of soft soil is very important on the elastic seismic response of a structure suggesting the necessity of the performance based seismic design.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.12 no.5
• /
• pp.397-404
• /
• 2002

In this paper, it is proposed wavelet neural network using the fuzzy concept with the fuzzy and the multi-resolution analysis(MRA) of wavelet transform. Also, it wishes to improve any nonlinear function learning approximation using this system. Here, the fuzzy concept is used the bell type fuzzy membership function. And the composition of wavelet has a unit size. It is used the backpropagation algorithm for learning of wavelet neural network using the fuzzy concept. It is used the multi-resolution analysis of wavelet transform, the bell type fuzzy membership function and the backpropagation algorithm for learning. This structure is confirmed to be improved approximation performance than the conventional algorithms from one dimension and two dimensions function through simulation.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.41 no.3
• /
• pp.1-8
• /
• 2004

The anomalous behavior of the 1/f noise of halo or pocket ion implanted MOSFETs is investigated. The model for the anomalous 1/f noise behaviors of MOSFETs, which consist of inhomogeneous conductance along the channel is improved within a regional approximation as previous works and presented in a fen directly applicable to halo MOSFETs. The presented model reduces to the previous results, discussed in the linear region operation, for small drain bias. Comparisons with experimental results show that the 1/f model based on the regional approach can be applicable for limited ranges, especially for sufficiently large gate bias voltages.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.27 no.4A
• /
• pp.324-329
• /
• 2002

In the high-speed/low-power digital filter applications like wireless communication systems, canonical signed digit(CSD) linear phase finite impulse response(FIR) filter structures are widely investigated. In this paper, we propose a high-speed/low-power CSD linear phase FIR filter structure using vertical common sub-expression. In the conventional linear phase CSD filter, horizontal common sub-expressions are utilized due to the inherent horizontal common sub-expression of symmetrical filter coefficients. We use the fact that their MSBs are also equal since adjacent filter coefficients have similar values in the linear phase filter Through the examples, it is shown that our proposed structure is more efficient in case that precision of implementation is lower, and tap length are longer.

• Journal of IKEEE
• /
• v.23 no.2
• /
• pp.716-721
• /
• 2019

A novel switching differentiator-based backstepping controller for uncertain pure-feedback nonlinear systems is proposed. Using asymptotically convergent switching differentiator, time-derivatives of the virtual controls are directly estimated in every backstepping design steps. As a result, the control law has an extremely simple form and asymptotical stability of the tracking error is guaranteed regardless of parametric or unstructured uncertainties and unmatched disturbances in the considered system. It is required no universal approximators such as neural networks or fuzzy logic systems that are adaptively tuned online to cope with system uncertainties. Simulation results show the simplicity and performance of the proposed controller.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.3
• /
• pp.747-753
• /
• 1997

Function approximation from a set of input-output parirs has numerous applications in scientiffc and engineer-ing areas.Multiayer feedforward neural networks have been proposed as a good approximator of noninear function.The back propagation (BP) algorithm allows muktiayer feedforward neural networks oro learn input-output mappongs from training samples.However, the mapping acquired through the BP algorithm nay be cor-rupt when errorneous trauning data are employed.In this paper we propose a robust BP learning algorithm that is resistant to the errormeous data and is capable of rejecting gross errors during the approximation process.