• 제어로봇시스템학회:학술대회논문집
• /
• 1991.10a
• /
• pp.168-173
• /
• 1991

A neural optimization network is proposed to control the redundant robot manipulators in an environment with the obstacle. The weightings of the network are adjusted by considering both the joint dexterity and the capability of collision avoidance of joint differential motion. The fuzzy rules are proposed to determine the capability of collision avoidance of each joint. To show the validities of the proposed method, computer simulation results are illustrated for the redundant robot of the planner type with three degrees of freedom.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2000.11a
• /
• pp.691-694
• /
• 2000

In this paper, a motion control algorithm is proposed by using neural network system, which makes a robot arm successfully avoid unexpected obstacle when the robot is moving from the start to the goal position. During the motion, if there is an obstacle the vision system recognizes it. And in every time the optimization-algorithm quickly chooses a motion among the possible motions of robot. The proposed algorithm has a good avoidance characteristic in simulation.

• 제어로봇시스템학회:학술대회논문집
• /
• 1992.10a
• /
• pp.490-495
• /
• 1992

A neural model predictive control strategy combining a neural network for plant identification and a nonlinear programming algorithm for solving nonlinear control problems is proposed. A constrained nonlinear optimization approach using successive quadratic programming cooperates with neural identification network is used to generate the optimum control law for the complicate continuous/batch chemical reactor systems that have inherent nonlinear dynamics. Based on our approach, we developed a neural model predictive controller(NMPC) which shows excellent performances on nonlinear, model-plant mismatch cases of chemical reactor systems.

• 제어로봇시스템학회:학술대회논문집
• /
• 1999.10a
• /
• pp.214-217
• /
• 1999

In this paper, genetic algorithm (GA) is implemented to search for the optimal structures (i.e. the kind of neural networks, the number of inputs and hidden neurons) of neural networks which are used approximating a given nonlinear function. Two kinds of neural networks, i.e. the multilayer feedforward [1] and time delay neural networks (TDNN) [2] are involved in this paper. The synapse weights of each neural network in each generation are obtained by associated training algorithms. The simulation results of nonlinear function approximation are given out and some improvements in the future are outlined.

• Proceedings of the KIEE Conference
• /
• 1995.11a
• /
• pp.55-58
• /
• 1995

An optimum design method of electric machines using neural network is presented. In this method, two multi - layer perceptrons of analysis and design neural network are used in optimizing process. A preliminary model of linear induction motor for subway is designed by the electric and magnetic loading distribution method and then optimized by presented method.

• Proceedings of the KIEE Conference
• /
• 2002.11d
• /
• pp.56-60
• /
• 2002

The paper presents the optimum design of a linear induction motor(LIM) using Genetic algorithm, Neural Network and SUMT. The design variables are optimized by three different optimization methods and the results are discussed.

• ETRI Journal
• /
• v.24 no.3
• /
• pp.251-254
• /
• 2002

We developed a method for optimal superframe design in the multi-frequency time division multiple access (MF-TDMA) return-link of a satellite multimedia interactive network called a digital video broadcasting return channel over satellite (DVB-RCS) sub-network. To find the optimal superframe pattern with the maximum data throughput, we formulated the design problem as a non-linear combinatorial optimization problem. We also devised the proposed simple method so that it would have field applicability for improving radio resource utilization in the MF-TDMA return link.

• ETRI Journal
• /
• v.14 no.3
• /
• pp.67-74
• /
• 1992

Discrete variable을 갖는 Mean Field Theory(MFT) neural network은 이미 많은 combinatorial optimization 문제에 적용되어져 왔다. 본 논문에서는 이를 확장하여 continuous variable을 갖는 mean field annealing을 제안하고, 이러한 network에서 integral로 표현되는 spin average를 mean field에 기초하여 어렵지 않게 구할 수 있는 one-variable stochastic simulated annealing을 제안하였다. 이런 방법으로 multi-body problem을 single-body problem으로 바꿀 수 있었다. 또한 이 방법을 이용한 응용으로서 통계학에서 잘 알려진 문제중의 하나인 quantification analysis 문제에 적용하여 타당성을 보였다.

• Journal of Communications and Networks
• /
• v.12 no.3
• /
• pp.246-252
• /
• 2010

A binary linear integer program is formulated for the problem of expanding the capacity of a fiber optic network and routing the traffic to deliver new interactive multimedia services. A two-phase Lagrangian dual search procedure and a Lagrangian heuristic are developed. Computational results show superior performance of the two-phase subgradient optimization compared with the conventional one-phase approach.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1993.06a
• /
• pp.899-902
• /
• 1993

A neural model predictive control strategy combining a neural network for plant identification and a nonlinear programming algorithm for solving nonlinear control problems is proposed. A constrained nonlinear optimization approach using successive quadratic programming combined with neural identification network is used to generate the optimum control law for complex continuous chemical reactor systems that have inherent nonlinear dynamics. The neural model predictive controller (MNPC) shows good performances and robustness. To whom all correspondence should be addressed.