• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2001.12a
• /
• pp.10-13
• /
• 2001

Nonlinear Function Approximation of Moduled Neural Network Using Genetic Algorithm Neural Network consists of neuron and synapse. Synapse memorize last pattern and study new pattern. When Neural Network learn new pattern, it tend to forget previously learned pattern. This phenomenon is called to catastrophic inference or catastrophic forgetting. To overcome this phenomenon, Neural Network must be modularized. In this paper, we propose Moduled Neural Network. Modular Neural Network consists of two Neural Network. Each Network individually study different pattern and their outputs is finally summed by net function. Sometimes Neural Network don't find global minimum, but find local minimum. To find global minimum we use Genetic Algorithm.

• Bulletin of the Korean Mathematical Society
• /
• v.50 no.3
• /
• pp.823-832
• /
• 2013

We present a new algorithm for solving a system of nonlinear equations with convex constraints which combines proximal point and projection methodologies. Compared with the existing projection methods for solving the problem, we use a different system of linear equations to obtain the proximal point; and moreover, at the step of getting next iterate, our projection way and projection region are also different. Based on the Armijo-type line search procedure, a new hyperplane is introduced. Using the separate property of hyperplane, the new algorithm is proved to be globally convergent under much weaker assumptions than monotone or more generally pseudomonotone. We study the convergence rate of the iterative sequence under very mild error bound conditions.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.29 no.8 s.239
• /
• pp.1051-1060
• /
• 2005

Nonlinear Response Optimization using Equivalent Static Loads (NROESL) method/algorithm is proposed to perform optimization of non-linear response structures. The conventional method spends most of the total design time on nonlinear analysis. The NROESL algorithm makes the equivalent static load cases for each response and repeatedly performs linear response optimization and uses them as multiple loading conditions. The equivalent static loads are defined as the loads in the linear analysis, which generates the same response field as those in non-linear analysis. The algorithm is validated for the convergence and the optimality. The proposed algorithm is applied to a simple mathematical problem to verify the convergence and the optimality.

• Journal of applied mathematics & informatics
• /
• v.28 no.3_4
• /
• pp.597-610
• /
• 2010

For a class of nonlinear bilevel programming problems in which the follower's problem is linear, the paper develops a genetic algorithm based on the optimality conditions of linear programming. At first, we denote an individual by selecting a base of the follower's linear programming, and use the optimality conditions given in the simplex method to denote the follower's solution functions. Then, the follower's problem and variables are replaced by these optimality conditions and the solution functions, which makes the original bilevel programming become a single-level one only including the leader's variables. At last, the single-level problem is solved by using some classical optimization techniques, and its objective value is regarded as the fitness of the individual. The numerical results illustrate that the proposed algorithm is efficient and stable.

• Proceedings of the KIEE Conference
• /
• 2003.11c
• /
• pp.580-583
• /
• 2003

The general time optimal control law provides the optimal solution for a minimum time control problem. But in most real systems with disturbances and model uncertainties, the time optimal control law leads to chattering effect. This chattering effect can cause the system to be unstable. Therefore, we propose a robust optimal control algorithm for the nonlinear second order systems with model uncertainty. The proposed algorithm is combined with bang-bang control and sliding mode control. Thus the proposed algorithm has two state space regions to implement to control algorithm. In each region, the appropriate linear or nonlinear feedback control law is used satisfying the dynamic system equations. Simulation results show the superiority of the proposed controller in comparison with pure time optimal control(bang-bang control).

• Communications for Statistical Applications and Methods
• /
• v.2 no.2
• /
• pp.135-145
• /
• 1995

A new algorithm was given to successively fit the multiexponential function/nonlinear function to data by a weighted least squares method, using Gauss-Newton, Marquardt, gradient and DUD methods for convergence. This study also considers the problem of linear-nonlimear weighted least squares estimation which is based upon the usual Taylor's formula process.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.28B no.12
• /
• pp.19-26
• /
• 1991

In this paper, we give a structure algorithm for the synthesis problems of the nonlinear system via dynamic feedback. Using our algorithm, sufficient conditions for the input-output synthesis problems are discussed. The problems we consider in this paper include dynamic input-output decoupling input-output linearization, and immersion into a linear system.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.49 no.10
• /
• pp.582-590
• /
• 2000

A fast optimal path planning algorithm using the analog Cellular Nonlinear Circuits(CNC) is proposed. The analog circuits based optimal path planning is very useful since most of the optimal path planning problems require real time computation. There has already been a previous study to implement the dynamic programming with analog circuits. However, it could not be applied for the practically large size of problems since the algorithm employs the mechanism of reducing its input current/voltage by the amount of cost, which causes outputs of distant cells to become zero. In this study, a subgoal-based dynamic programming algorithm to compute the optimal path is proposed. In the algorithm, the optimal paths are computed regardless of the distance between the starting and the goal points. It finds subgoals starting from the starting point when the output of the starting cell is raised from its initial value. The subgoal is set as the next initial position to find the next subgoal until the final goal is reached. The global optimality of the proposed algorithm is discussed and two different kinds of simulations have been done for the proposed algorithm.

• The Transactions of the Korean Institute of Electrical Engineers D
• /
• v.51 no.8
• /
• pp.344-351
• /
• 2002

The proposed immune algorithm has an uncomplicated structure and memory-cell mechanism as the optimization algorithm which imitates the principle of humoral immune response. We use the proposed algorithm to solve parameter optimization problems. Up to now, the applications of immune algorithm have been optimization problems with non-varying system parameters. Therefore the usefulness of memory-cell mechanism in immune algorithm is without. This paper proposes the immune algorithm using a memory-cell mechanism which can be the application of system with nonlinear varying parameters. To verified performance of the proposed immune algorithm, the speed control of nonlinear DC motor are performed. The results of Computer simulations represent that the proposed immune algorithm shows a fast convergence speed and a good control performances under the varying system parameters.

• Journal of the Korean Statistical Society
• /
• v.32 no.1
• /
• pp.1-10
• /
• 2003

We consider obtaining graphical summaries of uncertainty in estimates of parameters in nonlinear models. A nonlinear constrained optimization algorithm is developed for likelihood based confidence intervals for the functions of parameters in the model The results are applied to the problem of finding significance levels in nonlinear models.