• Journal of the Korea Society of Computer and Information
• /
• v.14 no.2
• /
• pp.27-35
• /
• 2009

Integer programming is a very effective technique for searching optimal solution of combinatorial optimization problems. However, its applicability is limited to linear models. In this paper, I propose an effective method for solving a nonlinear optimization problem by integrating the powerful search performance of integer programming and the flexibility of neighborhood search algorithms. In the first phase, integer programming is executed with subproblem which can be represented as a linear form from the given problem. In the second phase, a neighborhood search algorithm is executed with the whole problem by taking the result of the first phase as the initial solution. Through the experimental results using a nonlinear maximal covering problem, I confirmed that such a simple integration method can produce far better solutions than a neighborhood search algorithm alone. It is estimated that the success is primarily due to the powerful performance of integer programming.

• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.9
• /
• pp.371-380
• /
• 2022

Convolutional neural networks are widely used to manipulate data arranged in a grid, such as images. A general convolutional neural network consists of a convolutional layers and a fully connected layers, and each layer contains a nonlinear activation functions. This paper proposes a combined parametric activation function to improve the performance of convolutional neural networks. The combined parametric activation function is created by adding the parametric activation functions to which parameters that convert the scale and location of the activation function are applied. Various nonlinear intervals can be created according to parameters that convert multiple scales and locations, and parameters can be learned in the direction of minimizing the loss function calculated by the given input data. As a result of testing the performance of the convolutional neural network using the combined parametric activation function on the MNIST, Fashion MNIST, CIFAR10 and CIFAR100 classification problems, it was confirmed that it had better performance than other activation functions.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.18 no.2
• /
• pp.372-379
• /
• 1994

In this study, the describing function is adopted to represent nonlinearity in the system equations. The compliance can be obtained by solving nonlinear simultaneous algebraic quations for multi-degrees-of-freedom system with multinonlinearities. When the technique is applied, the nonlinearity of the system can be identified from the compliance which is obtained from the sinusoidal excitation of the system. By employing the describing function in the Building Block Analysis, we can extensively develop the BBA into investigation of the continuous systems with nonlinearities. The evaluated compliance can quantitatively show the effects of nonlinearity such as the transfer of the natural frequency, the variance of the compliance at the natural frequency, and the jump phenomena which occur during sweeping of the excitation frequency.

• Journal of Ocean Engineering and Technology
• /
• v.7 no.1
• /
• pp.73-80
• /
• 1993

본 연구에서는 장주형 해양구조물의 횡방향 진동에 대한 파라메트릭 가진 효과를 고찰하였다. 먼저, 장주형 해양구조물의 횡방향 운동에 대한 4계 편미방지배방정식을 비선형 Mathieu 방정식으로 유도하였다. 비선형 mathieu 방정식의 해를 구하여 장주형 해양구조물의 동적 반응 특성을 해석하였다. 유체 비선형 감쇠력은 불안정 조건하에 있는 파라메트릭 진동의 반응크기를 제한 하는데 중요한 역활을 한다. 파라메트릭 진동의 경우 가장 큰 반응크기는 Mathieu 안정차트의 첫번째 불안정 구간에서 일어난다. 반면에, 파라메트릭 진동과 강제진동의 결합 진동인 경우, 가장 큰 반응 크기는 두번째 불안정 구간에서 발생된다. 파라메트릭 가진으로 인한 장주형 해양구조물의 횡방향 운동은 동적조건에 따라 subharmonic, superharmonic 또는 chaotic 운동이 되기도 한다.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.44 no.4
• /
• pp.18-23
• /
• 2007

This paper presents a transmissive dispersion compensator based on tilted chirped fiber Bragg grating pairs which happen mode coupling between core mode and cladding mode. And, as a result of simulation, the phase matching condition and optimum tilted angle to maximize the mode-coupling in the dispersion compensator are shown and the dispersion slope and bandwidth in the proposed dispersion compensator is respectively 3,068ps/nm and 0.45nm.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.12 no.2
• /
• pp.251-264
• /
• 1999

본 연구에서는 추계론적 동적시스템의 응답거동을 예측할 수 있는 반해석적 절차를 개발하였으며, 이를 이용하여 구분적선형시스템의 동적거동특성을 확률적 영역에서 분석하였다. 반 해석적 절차는 시스템의 추계론적 미분방정식에 상응하는 Fokker-Planck 방정식을 path-integral solotion을 이용하여 풂으로써 구할 수 있다. 결합확률밀도함수의 시간에 따른 전개과정을 통하여 시스템의 동적 응답거동 특성의 예측과 분석을 하고 시스템의 거동에 미치는 외부노이즈의 영향 또한 조사하였다. 반 해석적 방법은 위상면 상에서 결합확률밀도 함수를 통하여 응답거동의 예측은 물론 거동특성에 대하여 적절한 정보를 제공하는 것을 밝혔다. 혼돈거동의 특성은 외부노이즈가 존재하는 상황에서도 시스템의 응답 안에 잔재하는 것을 밝혔다.

• Journal of the Korean Chemical Society
• /
• v.47 no.6
• /
• pp.667-670
• /
• 2003

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.41 no.7
• /
• pp.89-96
• /
• 2004

High-linearity voltage-controlled current sources (VCCSs) circuits for wide voltage-controlled oscillator and automatic gain control are proposed. The VCCS consists of emitter follower for voltage input, two common-base amplifier which their emitter connected for current output, and current mirror which connected the two amplifier for large output current. The VCCS used only five transistors and a resistor without an extra bias circuit. Simulation results show that the VCCS has current output range from 0㎃ to 300㎃ over the control voltage range from 1V to 4.8V at supply voltage 5V. The linearity error of output current has less than 1.4% over the current range from 0A to 300㎃.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.18 no.9
• /
• pp.1376-1385
• /
• 1993

This paper presents a new dynamic model which is represented by the second order differenatial equation and itcludes the robot arm dynamics as well as the actuator dynamics. The model exhibits excellent performance in the steady state and transient response. In addition the time varing nonlinear and coupled dynamic system has been linearized and decoupled by using nonlinear feedback and linearization method. In this case a pole assignment law is used to improve stability, and the optimal control altorithm is applied to the error equation to minimize the path error. In applying the proposed algorithm to the three joint manipulator with actuators, we obtained very encouraging results.

• Computational Structural Engineering
• /
• v.10 no.1
• /
• pp.213-223
• /
• 1997

In the analysis of the behavior of a complex structure, it requires much time and cost to analyze its behavior by using shell elements at the early design concept. For the purpose of the decrease of time and cost, many researches have been performed with the intention to analyze its behavior through replacing a shell model by a simple beam model. In the present study, a method is proposed for determining a bending spring stiffness which means the flexibility for applying into the joints of the simple beam model. Geometrically nonlinear analysis is performed through the application of the determined flexibility into joints of the simple beam model. The nonlinear behavior of thin-walled tube shell structure can be described within a little error through the simple beam model with flexible joints.