• The KIPS Transactions:PartA
• /
• v.8A no.3
• /
• pp.227-234
• /
• 2001

In this paper we propose a block-type parallel preconditioner for solving large sparse nonsymmetric linear systems, which we expect to be scalable. It is Multi-Color Block SOR preconditioner, combined with direct sparse matrix solver. For the Laplacian matrix the SOR method is known to have a nondeteriorating rate of convergence when used with Multi-Color ordering. Since most of the time is spent on the diagonal inversion, which is done on each processor, we expect it to be a good scalable preconditioner. We compared it with four other preconditioners, which are ILU(0)-wavefront ordering, ILU(0)-Multi-Color ordering, SPAI(SParse Approximate Inverse), and SSOR preconditiner. Experiments were conducted for the Finite Difference discretizations of two problems with various meshsizes varying up to $1025{\times}1024$. CRAY-T3E with 128 nodes was used. MPI library was used for interprocess communications, The results show that Multi-Color Block SOR is scalabl and gives the best performances.

• Journal of Intelligence and Information Systems
• /
• v.1 no.2
• /
• pp.103-121
• /
• 1995

This study proposes a artificial neural network method to predict the time to buy and sell the stocks listed on the Korea Composite Stock Price Index(KOSPI). Four types (NN1, NN2, NN3, NN4) of independent networks were developed to predict KOSPIs up/down direction after four weeks. These networks have a difference only in the length of learning period. NN5 - arithmetic average of four networks outputs - shows an higher accuracy than other network types and Multiple Linear Regression (MLR), and buying and selling simulation using systems outputs produces higher reture than buy-and-hold strategy.

• Journal of international Conference on Electrical Machines and Systems
• /
• v.3 no.2
• /
• pp.126-131
• /
• 2014

This paper discusses stability of new simplified sensorless vector control systems of induction motors (IM). The simplified sensorless systems estimate the flux angle by using the output voltage of d-axis PI current controller to achieve the q-axis flux zero. Two simplified sensorless systems are studied. The difference of two systems is the presence or absence of a q-axis PI current controller. The systems stability is compared by deriving linear state equations and showing root loci and unstable regions. Furthermore, transient responses and experiment results make clear the stability of the proposed system.

• Ocean and Polar Research
• /
• v.35 no.2
• /
• pp.147-155
• /
• 2013

The purpose of this study is to investigate seasonal difference in linear trends in satellite-derived chlorophyll-a concentration (Chl-a) and their related environmental changes in the South Sea of Korea (SSK) and East China Sea (ECS) for recent 15 years (Jan. 1998~Dec. 2012) by analyzing climatological data of Chl-a, Rrs(555), sea surface wind (SSW) and nutrient. A linear trend analysis of Chl-a data reveals that, during recent 15 years, the spring bloom was enhanced in most of the ECS, while summer and fall blooms were weakened. The increased spring (Mar. - May) Chl-a was associated with strengthened winter (Dec. - Feb.) wind that probably provided more nutrient into the upper ocean from the deep. The causes of decreased summer (Jun. - Aug.) Chl-a in the northern ECS were uncertain, but seemed to be related with the nutrient limitation. Recently (after 2006), low-salinity Changjiang diluted water in the south of Jeju and the SSK had lower phosphate that caused increase in N/P ratio with Chl-a decrease. The decreased fall (Sep. - Nov.) Chl-a was associated with weakened wind that tends to entrain less nutrient into the upper ocean from the deep. This study suggests that phytoplankton in the ECS differently changes in response to environmental changes depending on season and region.

• Earthquakes and Structures
• /
• v.14 no.3
• /
• pp.229-239
• /
• 2018

Under strong ground motion, pounding can be caused because of the different dynamic properties between two adjacent buildings. Using different structural systems in two adjacent structures makes a difference in the lateral stiffness and thus changes the pounding force between them. In this paper, the effect of the structural system of adjacent buildings on the amount of force applied by pounding effects has been investigated. Moment resisting frame systems (MRFs), lateral X-bracing system (LBS), shear wall system (SWS) and dual system (DS) have been investigated. Four different cases has been modelled using finite element (FE) method. The number of stories of the two adjacent buildings is different in each case: case 1 with 6 and 4 stories, case 2 with 9 and 6 stories, case 3 with 15 and 6 stories and case 4 with 10 and 10 stories. The structures have been modelled three-dimensionally. Non-linear time history analysis has been done on the structures using the finite element software SAP2000. In order to model pounding effects, the non-linear gap elements have been used.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.7
• /
• pp.3594-3607
• /
• 2017

The traditional feature extraction methods such as principal component analysis (PCA) cannot obtain the local structure of the samples, and locally linear embedding (LLE) cannot obtain the global structure of the samples. However, a common drawback of existing PCA and LLE algorithm is that they cannot deal well with the sparse problem of the samples. Therefore, by integrating the globality of PCA and the locality of LLE with a sparse constraint, we developed an improved and unsupervised difference algorithm called Sparse Difference Embedding (SDE), for dimensionality reduction of high-dimensional data in small sample size problems. Significantly differing from the existing PCA and LLE algorithms, SDE seeks to find a set of perfect projections that can not only impact the locality of intraclass and maximize the globality of interclass, but can also simultaneously use the Lasso regression to obtain a sparse transformation matrix. This characteristic makes SDE more intuitive and more powerful than PCA and LLE. At last, the proposed algorithm was estimated through experiments using the Yale and AR face image databases and the USPS handwriting digital databases. The experimental results show that SDE outperforms PCA LLE and UDP attributed to its sparse discriminating characteristics, which also indicates that the SDE is an effective method for face recognition.

• 제어로봇시스템학회:학술대회논문집
• /
• 2001.10a
• /
• pp.75.5-75
• /
• 2001

In this paper, we study design of a ILQ(Inverse Linear Quadratic optimal control) looper control system for hot strip mills. The looper which is placed between stands plays an important role in controlling strip width by regulating strip tension variation generated from the velocity difference of main work rolls. A Looper servo controller is designed by ILQ control theory which is an inverse problem of LQ(Linear Quadratic optimal control) control. The mathematical model for looper system is obtained by Taylor´s linearization of nonlinear differential equations. Then we designed linear controller for linearization model by using the ILQ control algorithm. Thereafter this controller is applied to the nonlinear model for model identification. As a result, we show the controller´s robustness for the model error, external disturbance and sensor noise.

• 제어로봇시스템학회:학술대회논문집
• /
• 1993.10a
• /
• pp.161-167
• /
• 1993

This paper presents robustness properties of the Kalman Filter ad the associated LQG/LTR method for linear time-invariant systems having delays in both the state and output. A circle condition relating to the return difference matrix associated with the Kalman filter is derived. Using this circle condition, it is shown that the Kalman filter guarantees(1/2, .inf.) gain margin and .+-.60.deg. phase margin, which are the same as those for nondelay systems. However, it is shown that, even for minimum phase plants, the LQG/LTR method can not recover the target loop transfer function. Instead, an upper bound on the recovery error is obtained using an upper bound of the solution of the Kalman filter Riccati equations. Finally, some dual properties between output-delated system and input-delayed systems are exploited.

• Journal of Institute of Control, Robotics and Systems
• /
• v.4 no.2
• /
• pp.141-150
• /
• 1998

This paper presents a dynamical anti-reset windup (ARW) compensation method for saturating control systems with multiple controllers and/or multiloop configuration. By regarding the difference of controller states in the absence and presence of saturating actuators as an objective function, the dynamical compensator which minimizes the objective function is derived in an integrated fashion. The proposed dynamical compensator is a closed form of plant and controller parameters. The resulting dynamics of compensated controller reflects the linear closed-loop system. The proposed method guarantees total stability of the resulting system. The effectiveness of the proposed method is illustrated by applying it to a servo motor control system. The paper is an extension of the results in Park and Choi[1].

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.25 no.5A
• /
• pp.618-633
• /
• 2000

In this paper, a Gaussian interpolation filter and cubic interpolation filter are presented to do more accurate channel estimation compared to the conventional linear interpolation filter for COFDM systems. In addition to an interpolation filter, a low pass filter using FFT and IFFT is also presented to reduce the noisy components of a channel estimate obtained by an interpolation filter. Channel estimates after low-pass filtering combined with interpolation filters can lower the error floor compared to the use of only interpolation filters. Computer simulation demonstrates that the presented channel estimation methods exhibit an improved performance compared to the conventional linear interpolation filter for COFDM systems in time-varying multipath fading channel and0.1 ~ 0.2 dB of Eb/No difference at BER=10-4 when the perfect channel estimation is compared.