• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2001년도 ICCAS
• /
• pp.43.1-43
• /
• 2001

We propose two control methods of the Lyapunov exponents for Jordan-type recurrent neural networks. Both the two methods are formulated by a gradient-based learning method. The first method is derived strictly from the definition of the Lyapunov exponents that are represented by the state transition of the recurrent networks. The first method can control the complete set of the exponents, called the Lyapunov spectrum, however, it is computationally expensive because of its inherent recursive way to calculate the changes of the network parameters. Also this recursive calculation causes an unstable control when, at least, one of the exponents is positive, such as the largest Lyapunov exponent in the recurrent networks with chaotic dynamics. To improve stability in the chaotic situation, we propose a non recursive formulation by approximating ...

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1989년도 하계종합학술대회 논문집
• /
• pp.679-684
• /
• 1989

In this paper, we describe the PTP notion of robot manipulators by neural networks. The PTP motion requires the inverse kinematic redline and the joint trajectory generation algorithm. We use the multi-layered Perceptron neural networks and the Error Back Propagation(EBP) learning rule for inverse kinematic problems. Varying the number of hidden layers and the neurons of each hidden layer, we investigate the performance of the neural networks. Increasing the number of learning sweeps, we also discuss the performance of the neural networks. We propose a method for solving the inverse kinematic problems by adding the error compensation neural networks(ECNN). And, we implement the neural networks proposed by Grossberg et al. for automatic trajectory generation and discuss the problems in detail. Applying the neural networks to the current trajectory generation problems, we can refute the computation time for trajectory generation.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2004년도 ICEIC The International Conference on Electronics Informations and Communications
• /
• pp.143-146
• /
• 2004

Differentiated services (DiffServ) has been widely accepted as the service model to adopt for providing quality-of­service (QoS) over the next-generation IP networks. There is a growing need to support QoS in mobile ad hoc networks. Supporting DiffServ in mobile ad hoc networks, however, is very difficult because of the dynamic nature of mobile ad hoc networks, which causes network congestion. The network congestion induces long transfer packet delay and low throughput which make it very difficult to support QoS in mobile ad hoc networks. We propose DiffServ module to support differentiated service in mobile ad hoc networks through congestion control. Our DiffServ module uses the periodical rate control for real time traffic and also uses the best effort bandwidth concession when network congestion occurs. Network congestion is detected by measuring the packet transfer delay or bandwidth threshold of real time traffic. We evaluate our mechanism via a simulation study. Simulation results show our mechanism may offer a low and stable delay and a stable throughput for real time traffic in mobile ad hoc networks.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2002년도 ICCAS
• /
• pp.106.5-106
• /
• 2002

1. Introduction 2. Artificial Immune Networks and Navigation Algorithm 3. Obstacle Avoidance and Goal Approach Behavior 4. Weights Adjustment Using Neural Network 5. Velocity Control and Local Minimum Avoidance 6. Simulation 7. Conclusion

• 제어로봇시스템학회논문지
• /
• 제9권4호
• /
• pp.329-339
• /
• 2003

A neural network based adaptive controller design method is proposed for reconfigurable flight control systems in the presence of variations in aerodynamic coefficients or control effectiveness decrease caused by control surface damage. The neural network based adaptive nonlinear controller is developed by making use of the backstepping technique for command following of the angle of attack, sideslip angle, and bank angle. On-line teaming neural networks are implemented to guarantee reconfigurability and robustness to the uncertainties caused by aerodynamic coefficients variations. The main feature of the proposed controller is that the adaptive controller is designed with assumption that not any of the nonlinear functions of the system is known accurately, whereas most of the previous works assume that only some of the nonlinear functions are unknown. Neural networks loam through the weight update rules that are derived from the Lyapunov control theory. The closed-loop stability of the error states is also investigated according to the Lyapunov theory. A nonlinear dynamic model of an F-16 aircraft is used to demonstrate the effectiveness of the proposed control law.

• 한국정밀공학회지
• /
• 제20권7호
• /
• pp.84-90
• /
• 2003

This paper develops a control method that is composed of the proportional control algorithm and the learning algorithm based on the recurrent neural networks (RNN) for the position control of a pneumatic rodless cylinder. The proportional control algorithm is suggested for the modeled pneumatic system, which is obtained easily simplifying the system, and the RNN is suggested for the compensation of the modeling errors and uncertainties of the pneumatic system. In the proportional control, two zones are suggested in the phase plane. One is the transient zone for the smooth tracking and the other is the small movement zone for the accurate position control with eliminating the stick-slip phenomenon. The RNN is connected in parallel with the proportional control for the compensation of modeling errors and frictions, compressibilities, and parameter uncertainties in the pneumatic control system. This paper experimentally verifies the feasibility of the proposed control algorithm for such pneumatic systems.

• 제어로봇시스템학회논문지
• /
• 제3권3호
• /
• pp.252-258
• /
• 1997

This paper presents neural networks for hybrid position/force control which is a type of position and force control for robot manipulators. The performance of conventional hybrid position/force control is excellent in the case of the exactly-known dynamic model of the robot, but degrades seriously as the uncertainty of the model increases. Hence, the neural network control scheme is presented here to overcome such shortcoming. The introduced neural term is designed to learn the uncertainty of the robot, and to control the robot through uncertainty compensation. Further more, the learning rule of the neural network is derived and is shown to be effective in the sense that it requires neither desired output of the network nor error back propagation through the plant. The proposed scheme is verified through the simulation of hybrid position/force control of a 6-dof robot manipulator.

• 전자공학회논문지SC
• /
• 제37권5호
• /
• pp.46-55
• /
• 2000

본 논문은 카오틱 신경망과 PD 제어기를 이용한 로봇 시스템의 직접적응제어 방식에 관한 것이다. 카오틱 신경망은 상·하층 결합계수 외에 궤환 결합계수와 동일 층 내의 결합계수를 가지며, 뉴런자체의 충분한 비선형성 때문에 강한 동적특성을 가지고 있다. 그러나 신경망의 구조 및 학습의 문제점으로 인하여 동적 시스템의 제어에 적용되지 못하고 있다. 본 논문에서는 기존의 카오틱 신경망을 제어 분야에 적용하기 위하여 적합한 구조로 수정하고 수정된 신경망의 학습에 관하여 고찰하였다. 제안된 신경망은 모의 실험을 통하여 3 축 푸마 로봇의 경로 제어에 적용하였다. 카오틱 신경망 제어기는 PD 제어기와 병렬로 구성하여 학습 초기의 안정성을 확보하였고, 제어대상의 비선형성을 보상하는 보상 제어기의 역할을 수행하도록 하였다

• 전자공학회논문지SC
• /
• 제37권3호
• /
• pp.72-83
• /
• 2000

본 논문에서는 리칭모드 제어기와 신경 회로망을 이용하여 3상 유도전동기의 위치제어 시스템을 구현한다. 리칭모드 제어기는 위치오차와 속도오차의 궤적을 슬라이딩 평면으로 들어가도록 하고 신경회로망의 초기학습을 담당한다. 리칭모드 제어기의 구조는 슬라이딩 평면의 스위칭 함수로부터 간단히 구성하였다. 또한, 신경 회로망은 전향경로 신경망으로 구성되며 비선형 매핑능력과 탁월한 학습능력을 이용하여 유도전동기의 등가제어입력을 학습하도록 하였고 신경 회로망의 입력으로는 모터의 기준속도, 기준위치 및 엔코더를 이용하여 측정된 모터의 실제속도와 위치 등을 이용하였고 온라인 상태로 학습되도록 하였다. 이와 같이 복합적으로 구성된 제어기들을 유도전동기의 위치제어 시스템에 적용하였다. 본 논문에서 제안된 알고리즘의 타당성을 보이기 위해 기존의 PI 제어기와 비교평가를 하였으며 시뮬레이션과 실험결과로부터 초기운전 상태에서는 리칭모드 제어기가 주로 제어를 담당하지만 시간이 지남에 따라 신경회로망이 학습되어 신경 회로망이 주 제어기가 됨을 확인하였다. 아울러, 제안된 제어기가 PI 제어기보다 우수하고 특히 부하변동과 같은 외란에 강인함을 알 수 있었으며, 정상상태 오차가 현저히 감소하여 정밀한 제어가 가능함을 확인하였다.

• 한국항공우주학회지
• /
• 제31권7호
• /
• pp.75-84
• /
• 2003

본 논문에서는 모델링 오차나 공력계수의 불확실성이 포함되어 있는 항공기 시스템에 대해서 신경회로망을 이용한 새로운 비선형 제어시스템 설계기법을 제안하였다. 비선형 적응제어법칙을 이용하여 신경회로망의 연결가중치를 변화시켰으며, 슬라이딩 제어법칙을 이용하여 신경회로망의 추정오차를 보상하였다. 제어시스템의 성능을 결정짓는 제어 매개변수들과 신경회로망 구조를 설계하기 위한 방법을 제안하였으며, 유전자 알고리듬을 이용하여 제어 매개변수들과 신경회로망 구조를 최적화하였따. 신경회로망의 구조탐색에 적합하도록 다수의 개체군을 형성하여 개체와 군이 동시에 전화하도록 하였다. 제안된 유전자 알고리듬에 의해 최적화된 구조를 갖는 신경회로망을 이용한 제어시스템을 항공기 종운동 모델에 적용하여 성능을 검증하였다.