• Journal of the Korea Society of Computer and Information
• /
• v.28 no.7
• /
• pp.11-18
• /
• 2023

In this paper, we evaluate deep learning time series forecasting models. Recent studies show that those models perform better than the traditional prediction model such as ARIMA. Among them, recurrent neural networks to store previous information in the hidden layer are one of the prediction models. In order to solve the gradient vanishing problem in the network, LSTM is used with small memory inside the recurrent neural network along with BI-LSTM in which the hidden layer is added in the reverse direction of the data flow. In this paper, we compared the performance of Informer by comparing with other models (LSTM, BI-LSTM, and Transformer) for real Nitrogen dioxide (NO2) data. In order to evaluate the accuracy of each method, mean square root error and mean absolute error between the real value and the predicted value were obtained. Consequently, Informer has improved prediction accuracy compared with other methods.

• Journal of the Korean Magnetics Society
• /
• v.13 no.1
• /
• pp.36-40
• /
• 2003

We have grown D $y_{x}$B $i_{3-x}$F $e_{5}$ $O_{12}$ (x = 0.5,1.0, 1.5,2.0) magnetic garnet thin films upon $Al_2$O3i and GGG substrate using Pechini process. The annealing temperature to get single phase D $y_{x}$B $i_{3-x}$F $e_{5}$ $O_{12}$ garnet is dependent on substrate, i.e. the annealing temperature for GGG substrate il 5$0^{\circ}C$ lower than that for $Al_2$ $O_3$ substrate. The grains of garnet thin film grown on GGG (111) plane align along [111] direction, and in this case the hysteresis curve does not saturate up to H : 5000 Oe. We attribute this phenomenon to rotation magnetization process. The maximum amount of Bi substitution in polycrystalline D $y_{x}$B $i_{3-x}$F $e_{5}$ $O_{12}$ thin film prepared by Pechini process is restricted to 2.0 Bi atom/unit cell, and this value is less than that in single garnet crystall grown by LPE method.own by LPE method.ethod.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.21 no.8
• /
• pp.880-885
• /
• 2010

In this paper, we propose a reconfigurable beam steering antenna using a dipole and a loop. The radiation patterns of the two antennas were cancelled or compensated, and head for the specific direction when a dipole and a loop antenna are combined at the reasonable ratio. The structure of the antenna is very simple and planar. By changing on/ off states of switches, the proposed antenna can steer the beam direction in the x-y plane. Simulation results confirmed the steering characteristic by using two imaginary switches. The proposed antenna can change the direction of the maximum gain in the x-y plane($0^{\circ}$, ${\pm}50^{\circ}$). The proposed antenna operates in 2.5~2.56 GHz(VSWR<2). It showed that peak gain of the antenna is 1.96~2.48 dBi and overall beam width of the reconfigurable antenna covers about $125^{\circ}$.

• Proceedings of the KSR Conference
• /
• 2007.05a
• /
• pp.99-108
• /
• 2007

The bi-modal tram guided by the magnetic guidance system has two car-bodies and three axles. Each axle of the vehicle has an independent suspension to lower the floor of the car and improve ride quality. The turning radius of the vehicle may increase as a consequence of the long wheel base. Therefore, the vehicle is equipped with the All-Wheel-Steering(AWS) system for safe driving on a curved road. Front and rear axles should be steered in opposite directions, which means a negative mode, to minimize the turning radius. On the other hand, they also should be steered in the same direction, which means a positive mode, for the stopping mode. Moreover, only the front axle is steered for stability of the vehicle upon high-speed driving. In summary, steering angles and directions of the each axle should be changed according to the driving environment and steering mode. This paper proposes an appropriate AWS control algorithm for stable driving of the bi-modal tram. Furthermore, a multi-body model of the vehicle is simulated to verify the suitability of the algorithm. This model can also analyze the different dynamic characteristics between 2WS and AWS.

• Transactions on Electrical and Electronic Materials
• /
• v.5 no.2
• /
• pp.55-60
• /
• 2004

High-temperature Superconductor(HTS) tubes were fabricated in terms of different chemical compositions and various SrSO$_4$additions by centrifugal forming method. For powder melting by induction the optimum range of melting temperatures and preheating temperature were 1050$^{\circ}C$, 1100$^{\circ}C$ and 550$^{\circ}C$ for 30 min, respectively. The mould rotating speed was 1000 rpm. A tube was annealed at 840$^{\circ}C$ for 72 hours in oxygen atmosphere. The plates like grains more than 20 $\mu\textrm{m}$ were well developed along the rotating direction of mould regardless of initial chemical compositions and the amounts of SrSO$_4$in Bi2212. The specimen with Bi2212 composition exhibited T$\_$c/ of 83 K, while the specimens fabricated with other compositions are lower than 60 K. The measured I$\_$c/ and J$\_$c/ at 77 K(B = 0 T) in Bi2212 with 7 % SrSO$_4$ composition were about 680 A and 380 A/$\textrm{cm}^2$.

• Proceedings of the Korean Geotechical Society Conference
• /
• 2009.09a
• /
• pp.884-890
• /
• 2009

The bi-directional pile load test with variable end plate overcomes the shortcoming of the Osterberg cell test. It is possible that the ultimate bearing capacity of the bi-direction can be known by using the loading of the end plate and two step procedures. The first step is to confirming end bearing capacity with variable end plate and the second step is similar to the conventional O-cell test. In the study, To calculate ultimate capacity of bi-directional load test using model with the pile with variable end plate O-cell, operated with end plate of 3 type on sand layer according to the relative density, loose, medium and dense conditions.

• Advances in nano research
• /
• v.8 no.1
• /
• pp.13-24
• /
• 2020

In this paper, for the first time based on the nonlocal strain gradient theory the effect of size dependency in torsional vibration of bi-direction functionally graded (FG) nonlinear nano-cone is study. The material properties were assumed to vary according to the arbitrary function in radial and axial directions. The Navier equation and boundary conditions of the size-dependent bidirectional FG nonlinear nano-cone were derived by Hamilton's principle. These equations were solved by employing the generalized differential quadrature method (GDQM). The presented model can turn into the classical model if the material length scale parameters are taken to be zero. The effects of some parameters, such as inhomogeneity constant, cross-sectional area parameter and small-scale parameters, were studied. As an essential result of this study can be stated that an FG nano-cone model based on the nonlocal elasticity theory behaves softer and based on the strain gradient theory behaves harder.

• The Transactions of the Korean Institute of Electrical Engineers C
• /
• v.49 no.3
• /
• pp.149-154
• /
• 2000

Bi-layer Mo films were deposited on soda-lime glass substrates using DC magnetron supttering. Increasing gas pressure, the resistivity varied from $1\times10^{-5}\; to\; 8.3\times10^{-3}\; \Omega.cm$. Furthermore, stress direction yielded compressive-to-tensile transition stress curves. The micro-structure of the compressively-stressed film which had poor adhesion consists of tightly packed columns, but of the tensile-stressed films had less dense structure. Under all gas pressure conditions, Mo films exhibited distinctly increasing optical reflection with decreasing gas pressure. The expansion of (110) peak width with the gas pressure meant the worse crystalline growth. The impurity levels in the Mo film exhibited highly concentrated Na, Se and O elements due to less dense micro-structure. The degree of Na diffusion depends on the type of the glass substrate used and the nature of the Mo film.

• Management Science and Financial Engineering
• /
• v.13 no.2
• /
• pp.1-27
• /
• 2007

This paper presents a fuzzy-goal programming(FGP) approach for Bi-Level Linear Multiple Objective Decision Making(BLL-MODM) problem in a large hierarchical decision making and planning organization. The proposed approach combines the attractive features of both fuzzy set theory and goal programming(GP) for MODM problem. The GP problem has been developed by fixing the weights and aspiration levels for generating pareto-optimal(satisfactory) solution at each level for BLL-MODM problem. The higher level decision maker(HLDM) provides the preferred values of decision vector under his control and bounds of his objective function to direct the lower level decision maker(LLDM) to search for his solution in the right direction. Illustrative numerical example is provided to demonstrate the proposed approach.

• East Asian Economic Review
• /
• v.19 no.1
• /
• pp.39-70
• /
• 2015

This paper examines the effects of foreign direct investment (FDI) on South Asian economies' output growth, utilizing recent panel cointegration testing and estimation techniques. Annual panel data on eight SAARC (South Asian Association for Regional Cooperation) member countries' macroeconomic variables over the period 1960- 2013 are employed in empirical analysis. Using various heterogeneous panel cointegration and panel causality tests, a bi-directional relationship between FDI and growth is found. We find evidence for both FDI-led growth and growth-induced FDI hypotheses for the South Asian economies over the sample period. Individual member countries exhibit heterogeneity in terms of the direction or existence of causality subject to their idiosyncratic economic conditions. Among various regressors, FDI, financial development, human capital, and government consumption show the most significant positive effects on output growth. As determinants of FDI, GDP, financial development, human capital, and government consumption are found significant in the region. The bi-directional causality between FDI and growth is found robust to the inclusion of other control variables and using different estimation techniques.