• The Transactions of the Korean Institute of Electrical Engineers P
• /
• v.65 no.2
• /
• pp.108-113
• /
• 2016

The real-world problems usually show nonlinear and multi-variate characteristics, so it is difficult to establish concrete mathematical models for them. Thus, it is common to practice data-driven modeling techniques in these cases. Among them, most widely adopted techniques are regression model and intelligent model such as neural networks. Regression model has drawback showing lower performance when much non-linearity exists between input and output data. Intelligent model has been shown its superiority to the linear model due to ability capable of effectively estimate desired output in cases of both linear and nonlinear problem. This paper proposes modeling method of daily photovoltaic power systems using ELM(Extreme Learning Machine) based modular networks. The proposed method uses sub-model by fuzzy clustering rather than using a single model. Each sub-model is implemented by ELM. To show the effectiveness of the proposed method, we performed various experiments by dataset acquired during 2014 in real-plant.

• Management Science and Financial Engineering
• /
• v.15 no.1
• /
• pp.51-69
• /
• 2009

A network model and a Genetic Algorithm (GA) is proposed to solve the simultaneous estimation of the trip distribution and traffic assignment from traffic counts in the congested networks in a logit-based Stochastic User Equilibrium (SUE). The model is formulated as a problem of minimizing a non-linear objective function with the linear constraints. In the model, the flow-conservation constraints are utilized to restrict the solution space and to force the link flows become consistent to the traffic counts. The objective of the model is to minimize the discrepancies between two sets of link flows. One is the set of link flows satisfying the constraints of flow-conservation, trip production from origin, trip attraction to destination and traffic counts at observed links. The other is the set of link flows those are estimated through the trip distribution and traffic assignment using the path flow estimator in the logit-based SUE. In the proposed GA, a chromosome is defined as a real vector representing a set of Origin-Destination Matrix (ODM), link flows and route-choice dispersion coefficient. Each chromosome is evaluated by the corresponding discrepancies. The population of the chromosome is evolved by the concurrent simplex crossover and random mutation. To maintain the feasibility of solutions, a bounded vector shipment technique is used during the crossover and mutation.

• The Bulletin of The Korean Astronomical Society
• /
• v.39 no.1
• /
• pp.37.2-37.2
• /
• 2014

Gas in disk galaxies interacts nonlinearly with a underlying stellar spiral potential to form galactic spiral shocks. Numerical simulations typically show that these shocks are unstable to the wiggle instability, forming non-axisymmetric structures with high vorticity. While previous studies suggested that the wiggle instability may arise from the Kelvin-Helmholtz instability or orbit crowding of gas elements near the shock, its physical nature remains uncertain. It was even argued that the wiggle instability is of numerical origin, caused by the inability of a numerical code to resolve a shock that is inclined to numerical grids. In this work, we perform a normal-mode linear stability analysis of galactic spiral shocks as a boundary-value problem. We find that the wiggle instability originates physically from the potential vorticity generation at a distorted shock front. As the gas follows galaxy rotation, it periodically passes through multiple shocks, successively increasing its potential vorticity. This sets up a normal-mode that grows exponentially, with a growth rate comparable to the orbital angular frequency. We show that the results of our linear stability analysis are in good agreement with the those of local hydrodynamic simulations of the wiggle instability.

• Structural Engineering and Mechanics
• /
• v.62 no.4
• /
• pp.401-415
• /
• 2017

A new higher shear deformation theory (HSDT) is presented for the thermal buckling behavior of functionally graded (FG) sandwich plates. It uses only four unknowns, which is even less than the first shear deformation theory (FSDT) and the conventional HSDTs. The theory considers a hyperbolic variation of transverse shear stress, respects the traction free boundary conditions and contrary to the conventional HSDTs, the present one presents a new displacement field which includes undetermined integral terms. Material characteristics and thermal expansion coefficient of the sandwich plate faces are considered to be graded in the thickness direction according to a simple power-law distribution in terms of the volume fractions of the constituents. The core layer is still homogeneous and made of an isotropic material. The thermal loads are supposed as uniform, linear and non-linear temperature rises within the thickness direction. An energy based variational principle is used to derive the governing equations as an eigenvalue problem. The validation of the present work is carried out with the available results in the literature. Numerical results are presented to demonstrate the influences of variations of volume fraction index, length-thickness ratio, loading type and functionally graded layers thickness on nondimensional thermal buckling loads.

• International Journal of Computer Science & Network Security
• /
• v.23 no.11
• /
• pp.21-31
• /
• 2023

In this paper, we focused on the problem of evaluating multi-class classification accuracy and simulation of multiple classifier performance metrics. Multi-class classifiers for sentiment analysis involved many challenges, whereas previous research narrowed to the binary classification model since it provides higher accuracy when dealing with text data. Thus, we take inspiration from the non-linear Support Vector Machine to modify the algorithm by embedding dynamic hyperplanes representing multiple class labels. Then we analyzed the performance of multi-class classifiers using macro-accuracy, micro-accuracy and several other metrics to justify the significance of our algorithm enhancement. Furthermore, we hybridized Enhanced Convolution Neural Network (ECNN) with Dynamic Support Vector Machine (DSVM) to demonstrate the effectiveness and efficiency of the classifier towards multi-class text data. We performed experiments on three hybrid classifiers, which are ECNN with Binary SVM (ECNN-BSVM), and ECNN with linear Multi-Class SVM (ECNN-MCSVM) and our proposed algorithm (ECNNDSVM). Comparative experiments of hybrid algorithms yielded 85.12 % for single metric accuracy; 86.95 % for multiple metrics on average. As for our modified algorithm of the ECNN-DSVM classifier, we reached 98.29 % micro-accuracy results with an f-score value of 98 % at most. For the future direction of this research, we are aiming for hyperplane optimization analysis.

• Geomechanics and Engineering
• /
• v.11 no.2
• /
• pp.235-252
• /
• 2016

This paper presents a new numerical solution of the ultimate lateral capacity of rectangular piles in clay. The two-dimensional plane strain finite element was employed to determine the limit load of this problem. A rectangular pile is subjected to purely lateral loading along either its major or minor axes. Complete parametric studies were performed for two dimensionless variables including: (1) the aspect ratios of rectangular piles were studied in the full range from plates to square piles loaded along either their major or minor axes; and (2) the adhesion factors between the soil-pile interface were studied in the complete range from smooth surfaces to rough surfaces. It was found that the dimensionless load factor of rectangular piles showed a highly non-linear function with the aspect ratio of piles and a slightly non-linear function with the adhesion factor at the soil-pile interface. In addition, the dimensionless load factor of rectangular piles loaded along the major axis was significantly higher than that loaded along the minor axis until it converged to the same value at square piles. The solutions of finite element analyses were verified with the finite element limit analysis for selected cases. The empirical equation of the dimensionless load factor of rectangular piles was also proposed based on the data of finite element analysis. Because of the plane strain condition of the top view section, results can be only applied to the full-flow failure mechanism around the pile for the prediction of limiting pressure at the deeper length of a very long pile with full tension interface that does not allow any separation at soil-pile interfaces.

• Journal of Advanced Navigation Technology
• /
• v.20 no.1
• /
• pp.53-58
• /
• 2016

In this paper, we propose a novel quadratic Kalman filter based object tracking algorithm using moving pictures. Quadratic Kalman filter, which is introduced recently, has not yet been applied to the problem of 3-dimensional (3-D) object tracking. Since the mapping of a position in 2-D moving pictures into a 3-D world involves non-linear transformation, appropriate algorithm must be chosen for object tracking. In this situation, the quadratic Kalman filter can achieve better accuracy than extended Kalman filter. Under the same conditions, we compare extended Kalman filter, unscented Kalman filter and sequential importance resampling particle filter together with the proposed scheme. In conculsion, the proposed scheme decreases the divergence rate by half compared with the scheme based on extended Kalman filter and improves the accuracy by about 1% in comparison with the one based on unscented Kalman filter.

• Journal of the Korea Society of Computer and Information
• /
• v.18 no.7
• /
• pp.37-45
• /
• 2013

Context dependent fusion (CDF) is a fusion algorithm that combines multiple outputs from different classifiers to achieve better performance. CDF tries to divide the problem context into several homogeneous sub-contexts and to fuse data locally with respect to each sub-context. CDF showed better performance than existing methods, however, it is sensitive to noise due to the large number of parameters optimized and the innate linearity limits the application of CDF. In this paper, a variant of CDF using support vector machines (SVMs) for fusion and kernel principal component analysis (K-PCA) for context extraction is proposed to solve the problems in CDF, named CDF-SVM. Kernel PCA can shape irregular clusters including elliptical ones through the non-linear kernel transformation and SVM can draw a non-linear decision boundary. Regularization terms is also included in the objective function of CDF-SVM to mitigate the noise sensitivity in CDF. CDF-SVM showed better performance than CDF and its variants, which is demonstrated through the experiments with a landmine data set.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.4
• /
• pp.688-697
• /
• 2017

It is very difficult to solve mathematically the optimal control problem for non - linear mobile robots to move to target points with minimum energy related to velocity, acceleration and angular velocity in minimum time. This paper proposes a method to obtain optimal control gains with which mobile robots move with minimum energy related to velocity, acceleration and angular velocity in minimum time using genetic algorithms. Mobile robots are non - linear systems so that their optimal control gains depend on initial positions. Hence initial positions are divided into some partition points and optimal control gains are obtained at each partition point with genetical algorithms. These optimal control gains are used to train neural networks that generate proper control gains at arbitrary initial position. Finally computer simulation studies have been conducted to verify the effectiveness of the method proposed in this paper.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.05a
• /
• pp.681-684
• /
• 2015

Power amplifier is an essential component for transmitting signals to a remote receiver in wireless communication systems. Power amplifier is a non-linear device in general, and the nonlinear distortion becomes severer as the output power increases. The nonlinearity results in spectral regrowth, which leads to adjacent channel interference, and decreases the transmit signal quality. To linearize power amplifiers, many techniques have been developed so far. Among the techniques, digital pre-distortion is known as the most cost and performance effective technique. However, the linearization performance falls down abruptly when the power amplifier operates in its saturation region. This is because of the severe nonlinearity. To relieve this problem, this paper proposes a new adaptive predistortion technique. The proposed technique controls the adaptive algorithm based on the power amplifier input level. Specifically, for small signals, the adaptive predistortion algorithm works normally. On the contrary, for large signals, the adaptive algorithm stops until small signals occur again. By doing this, wrong coefficient update by severe nonlinearity can be avoided. Computer simulation results show that the proposed method can improve the linearization performance compared with the conventional digital predistortion algorithms.