• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제24권4호
• /
• pp.331-350
• /
• 2020

Exploding gradient is a widely known problem in training recurrent neural networks. The explosion problem has often been coped with cutting off the gradient norm by some fixed value. However, this strategy, commonly referred to norm clipping, is an ad hoc approach to attenuate the explosion. In this research, we opt to view the problem from a different perspective, the discrete-time optimal control with infinite horizon for a better understanding of the problem. Through this perspective, we fathom the region at which gradient explosion occurs. Based on the analysis, we introduce a gradient-explosion-free algorithm that keeps the training process away from the region. Numerical tests show that this algorithm is at least three times faster than the clipping strategy.

• Steel and Composite Structures
• /
• 제31권6호
• /
• pp.641-653
• /
• 2019

In this paper, wave propagation is studied and analyzed in double-layered nanotubes systems via the nonlocal strain gradient theory. To the author's knowledge, the present paper is the first to investigate the wave propagation characteristics of double-layered porous nanotubes systems. It is generally considered that the material properties of nanotubes are related to the porosity and hygro-thermal effects. The governing equations of the double-layered nanotubes systems are derived by using the Hamilton principle. The dispersion relations and displacement fields of wave propagation in the double nanotubes systems which experience three different types of motion are obtained and discussed. The results show that the phase velocities of the double nanotubes systems depend on porosity, humidity change, temperature change, material composition, non-local parameter, strain gradient parameter, interlayer spring, and wave number.

• 한국ITS학회 논문지
• /
• 제13권4호
• /
• pp.48-56
• /
• 2014

유류소모량 산정 모형 개발을 위해서는 차종, 도로의 경사, 포장상태, 포장종류 등 다양한 변수들을 고려해야 하지만 현재 사용하고 있는 국토교통부의 투자평가지침에도 차종으로만 구분이 되어 있을 뿐 다양한 요인들을 고려하지 못하고 있는 실정이다. 본 연구에서는 도로의 경사도가 승용차의 유류소모량에 미치는 영향을 분석하기 위해 실제 주행실험을 통해 얻은 데이터를 기반으로 유류소모량 산정 모형을 개발하고 적용성을 검증하는 것을 목적으로 한다. 경사도에 따른 유류소모량 모형 개발을 위해 GPS 장비와 연비측정장비를 이용하여 실제 주행실험을 통해 유류소모량을 초(sec)단위로 측정하였다. 평지(${\pm}0{\sim}2%$), 오르막(+2~5%), 내리막(-2~5%)의 세 가지 경사도로 구분하였으며 차량의 속도와 유류소모량을 변수로 하는 회귀모형을 이용하여 모형을 개발하였다. 승용차의 유류소모량은 내리막, 평지, 오르막 순으로 커지는 것을 확인할 수 있었다.

• Journal of Astronomy and Space Sciences
• /
• 제16권2호
• /
• pp.91-104
• /
• 1999

We use BV CCD images to investigate the reality of the color gradient within a King type globular cluster NGC 7089. Surface photometry shows that there is a strong radial color gradient in the central region of the cluster in the sense of bluer center with the amplitude of ~0.39 $\pm$ 0.07 mag/$arcsec^2$ in (B - V). In the outer region of the cluster, however, the radial color gradient shows a reverse case, i.e., redder toward the center. (B - V) color profile which was derived from resolved stars in VGC 7089 field also shows a significant color gradient in the central region of the clusters, indicating that lights from the combination of red giant stars and blue horizontal branch stars cause the radial color gradient. Color gradient of the outer region of NGC 7089 may be due to the unresolved background of the cluster. Similar color gradients in the central area of clusters have been previously observed exserved exclusively in highly concentrated systems classified as post core collapse clusters. We caution, however, to confirm the reality of the color gradient from resolved stars, we need more accurate imaging data of the cluster with exceptional seeing condition because the effect of completeness correlates with local density of stars.

• International Journal of Control, Automation, and Systems
• /
• 제3권1호
• /
• pp.43-55
• /
• 2005

In this paper, a predictive control method using self-recurrent wavelet neural network (SRWNN) is proposed for chaotic systems. Since the SRWNN has a self-recurrent mother wavelet layer, it can well attract the complex nonlinear system though the SRWNN has less mother wavelet nodes than the wavelet neural network (WNN). Thus, the SRWNN is used as a model predictor for predicting the dynamic property of chaotic systems. The gradient descent method with the adaptive learning rates is applied to train the parameters of the SRWNN based predictor and controller. The adaptive learning rates are derived from the discrete Lyapunov stability theorem, which are used to guarantee the convergence of the predictive controller. Finally, the chaotic systems are provided to demonstrate the effectiveness of the proposed control strategy.

• 한국지능시스템학회논문지
• /
• 제14권1호
• /
• pp.118-124
• /
• 2004

In this paper, we present a indirect adaptive control method using a wavelet neural network (WNN) for the control of chaotic nonlinear systems without precise mathematical models. The proposed indirect adaptive control method includes the off-line identification and on-line control procedure for chaotic nonlinear systems. In the off-line identification procedure, the WNN based identification model identifies the chaotic nonlinear system by using the serial-parallel identification structure and is trained by the gradient-descent method. And, in the on-line control procedure, a WNN controller is designed by using the off-line identification model and is trained by the error back-propagation algorithm. Finally, the effectiveness and feasibility of the proposed control method is demonstrated with applications to the chaotic nonlinear systems.

• Journal of Advanced Marine Engineering and Technology
• /
• 제25권5호
• /
• pp.1115-1121
• /
• 2001

This paper presents the development of a tracking algorithm for shipborad satellite antenna systems which can enhance the tracking performance. In order to overcome some drawbacks of the conventional step tracking algorithm a new tracking algorithm is proposed. The proposed algorithm searches for the best tracking angles using gradient-based formulae and signal intensities measured according to a search pattern. The effectiveness of the proposed algorithm is demonstrated through simulation using real data.

• 한국항해학회지
• /
• 제25권4호
• /
• pp.361-369
• /
• 2001

This thesis describes the design of a stabilized algorithm for shipboard satellite antenna systems which can enhance the tracking performance. In order to overcome some drawbacks of the conventional step tracking algorithm, the proposed algorithm searches for the best tracking angles using gradient-based formulae and signal intensities measured according to a search pattern. The effectiveness of the proposed algorithm is demonstrated through simulation using real-world data.

• 충청수학회지
• /
• 제20권1호
• /
• pp.65-70
• /
• 2007

We study the existence of nontrivial solutions of the Gradient type Dirichlet boundary value problem for elliptic systems of the form $-{\Delta}U(x)={\nabla}F(x,U(x)),x{\in}{\Omega}$, where ${\Omega}{\subset}R^N(N{\geq}1)$ is a bounded regular domain and U = (u, v) : ${\Omega}{\rightarrow}R^2$. To study the system we use the liking theorem on product space.

• 한국마린엔지니어링학회:학술대회논문집
• /
• 한국마린엔지니어링학회 2001년도 춘계학술대회 논문집
• /
• pp.219-224
• /
• 2001

This paper presents the development of a tracking algorithm for shipboard satellite antenna systems which can enhance the tracking performance. In order to overcome some drawbacks of the conventional step tracking algorithm, the proposed algorithm searches for the best tracking angles using gradient-based formulae and signal intensities measured according to a search pattern. The effectiveness of the proposed algorithm is demonstrated through simulation using real-world data.