• 전자공학회논문지B
• /
• 제31B권5호
• /
• pp.123-133
• /
• 1994

This paper presents an output arbitrator for input buffering ATM (Asynchronous Transfer Mode) switches using neural networks. To avoid blocking in ATM switches with blocking characteristics, it is required to buffer ATM cells in input buffer and to schedule them. The N$\times$N request matrix is divided into N/16 submatrices in order to get rid of internal blocking systematically in scheduling phase. The submatrices are grouped into N/4 groups, and the cells in each group are switched alternatively. As the window size of input buffer is increases, the number of input cells switched in a time slot approaches to N. The selection of nonblocking cells to be switched is done by neural network modules. N/4 neural network modules are operated simultaneously. Fast selection can be achieved by massive parallelism of neural networks. The neural networks have 4N neurons and 14N connection. The proposed method is implemented in C language, and the simulation result confirms the feasibility of this method.

• 전자공학회논문지A
• /
• 제31A권10호
• /
• pp.161-172
• /
• 1994

This paper describes the design of a neurochip with on-chip learning capability in hardware with multiple processing elements. A digital architecture is adopted because its flexiblity and accuracy is advantageous for simulating the various application systems. The proposed chip consists of several processing elements to fit the large computation of neural networks, and has on-chip learning capability based on error back-propagation algorithm. It also minimizes the number of blobal control signals for processing elements. The modularity of the system makes it possible to buil various kinds of boards to match the expected range of applications.

• 전자공학회논문지B
• /
• 제33B권2호
• /
• pp.140-148
• /
• 1996

This paper proposes a combined feedforward/feedback controller which uses jacobians of neural network. The jacobians are calculated form the neural network that identifies the nonlinear plant, which are used for designing a jacobian controller and for training a neural network controller. Normally, it takes much time to train the neural network controller. Combining the neural and the jacobian controller, it can be a stable controller from the beginning of training phase of neural network, and it can be implemented as a learning-while-functioning controller. Simulated resutls for the proposed controller show its effectiveness and better performances.

• 전자공학회논문지B
• /
• 제32B권12호
• /
• pp.1652-1662
• /
• 1995

In this paper, a hybrid neural network is proposed to improve the learning ability of a neural network. The union of the characteristics of a Self-Organizing Neural Network model and of multi-layer perceptron model using the backpropagation learning method gives us the advantage of reduction of the learning error and the learning time. In learning process, the proposed hybrid neural network reduces the number of nodes in hidden layers to reduce the calculation time. And this proposed neural network uses the fuzzy feedback values, when it updates the responding region of each node in the hidden layer. To show the effectiveness of this proposed hybrid neural network, the boolean function(XOR, 3Bit Parity) and the solution of inverse kinematics are used. Finally, this proposed hybrid neural network is applied to the visual tracking control of a PUMA560 robot, and the result data is presented.

• 전자공학회논문지C
• /
• 제35C권8호
• /
• pp.68-74
• /
• 1998

The neural network predictiv econtroller (NNPC) is proposed for the attempt to mimic the function of brain that forecasts the future. It consists of two loops, one is for the prediction of output (NNP:neural network predictor) and the other one is for control the plant(NNC: neural network controller). The output of NNC makes the control input of plant, which is followed by the variation of both plant error and predictin error. The NNP forecasts the future output based upon the current control input and the estimated control output. The input and the output data of a system and a new method using evolution strategy are used to train the NNP. A two-step NNPC is applied to control the temeprature in boiler systems. It was compared with PI controller and auto-tuning PID controller. The computer simulaton and experimental results show that the proposed method has better performances than the other method.

• 전자공학회논문지S
• /
• 제34S권9호
• /
• pp.41-49
• /
• 1997

In this paper, we describe the algorithm which controls an unknown nonlinear system with disturbance a using multilayer neural network. The multilayer neural network can be used to approximate any continuous function to any desired degree of accuracy. With the former fact, we approximate an unknown nonlinear system by using of multilayer neural netowrk. WE include a disturbance among the modelling error, and the weight-update rule of multilayer neural network is derived to satisfy Laypunov stability. The whole control system constitutes controller using the feedback linearization method. The weight of neural network which is used to implement nonlinear function is updated by the derived update-rule. The proposed control algorithm is verified through computer simulation.

• 전자공학회논문지B
• /
• 제33B권8호
• /
• pp.124-133
• /
• 1996

This paper presents the implementation of variable structure control system for a linear or nonlinear system using neural networks. The overall control system consists of neural network controller and a reaching mode controller. While the former approximates the equivalent control input on the sliding surface, the latter is used to bring the entire system trajectories toward the sliding surface. No supervised learning procedures are needed and the weights of the neural network are tuned on-line automatically. The neural netowrk-based variable structure control system is applied to a nonlinare unstable inverted pendulum system through computer simulations, and implemented using a microcomputer (80486-50MHz) and applied to the DC servomotor position control system. Simulation and experimental results show the expected approximation sliding property is occurred. The proposed controller is compared with a PID controller and shows better performance than the PID controller in abrupt plant parameter change.

• 전자공학회논문지B
• /
• 제33B권7호
• /
• pp.32-39
• /
• 1996

In this paper, we describe the algorithm which controls an unknown nonlinear system with multilayer neural network. The multilayer neural netowrk can be used to approximate any continuous function to any desired degree of accuracy. With the former fact, we approximate unknown nonlinear function on the nonlinear system by using of multilayer neural netowrk. The weights on the hidden layer of multilayer neural network are updated by gradient method. The weight-update rule on the output layer is derived to satisfy lyapunov stability. Also, we obtain secondary controller form deriving step. The global control system consists of controller using feedback linearization method and secondary controller is order to satisfy layapunov stability. The proposed control algorithm is verified through computer simulation.

• 조명전기설비학회논문지
• /
• 제18권6호
• /
• pp.37-44
• /
• 2004

전력계통분야의 복합 대형화에 유연한 대처와 전력조류의 최적화 도모를 위해 사용되는 FACTS(Flexible AC Transmission System)기기 중 가장 유용한 UPFC(Unified Power Flow Controller)는 선로의 전압을 임의의 크기와 위상을 갖도록 제어하여 선로로 전송되는 유ㆍ무효전력을 총체적으로 보상하는 기능을 갖는다. 이런 UPEC가 계통에 연계되어 운영된다면 송전선로 매개변수가 변하기 때문에 계통의 영향을 많이 받는 거리계전기는 불필요한 오동작이 발생하게 된다. 즉 거리계전기에서 바라본 임피던스 영역(Impedance Zone)이 송전선로에 UPFC 연계시 각각의 보상 값에 의해 상당한 변화를 보임으로, 기존의 방식으로 정정된 Relay Setting Zone과 Adaptive Setting Zone은 현저한 오차가 발생하게 된다. 그러므로 계통에 연계된 UPFC의 운전 조건을 고려한 거리계전기 보호구간의 재설정이 필요하게 된다. 따라서 본 논문의 목적은 학습이 가능한 신경회로망(ANN)을 이용하여 거리계전기 동작의 신속성(Speed)을 기본으로 전력계통의 다양한 환경에 대해 거리계전기 응동 특성을 향상시키는데 있다. 학습 방법으로는 정적 및 동적인 비선형 시스템의 인식과 다변수 시스템에 적용 가능한 역전파 알고리즘(Back-propagation Algorithm)을 사용했다.

• 드라이브 ㆍ 컨트롤
• /
• 제13권2호
• /
• pp.34-41
• /
• 2016

Unlike the commonly used On/Off solenoid, constant attraction force which is independent of plunger displacement is a considerably important characteristic to proportional solenoid of the EPPR Valve. Attraction force uniformity is mainly affected by the internal shape design parameters. Due to a number of shape design parameters, the optimal parameter values are very complex and time consuming to find by trial and error method. Much research has been conducted or are still in progress to find the optimal parameter values by applying various optimization techniques like Genetic Algorithm, Evolution Strategy, Simulated Annealing, or the Taguchi method. In this paper, the design parameters which have primary effects on the attraction force uniformity and the average attraction force are decided by main effects analysis of Design of Experiments. Optimal parameter values are derived using finite-element analysis and a neural network model.