비선형 시스템의 직접제어방식을 위한 다층 신경회로망

이현관 호남대학교 2004 호남대학교 학술논문집 Vol.25 No.2

In this paper, we proposed a multi-layer neural network for direct control method of nonlinear system. The proposed control method uses neural network as the controller to learn inverse model of plant. The neural network used consists of two parts; one part is for identification of linear part, and the other is for identification of nonlinear part of inverse system. The neural network has to be learned the liner part with RLS algorithm and the nonlinear part with error of plant. From the simulation and experiment of tracking control to use one link manipulator as plant, we proved usefulness of the proposed control method to comparing to conventional direct neural network control method. By comparing the two methods, from simulation and experiment, we were convinced that the proposed control method is more simple and accuracy than the conventional method. Moreover, number of weight and bias to be controller parameter are small, and it has smaller steady state error than conventional method.

전방향 신경회로망을 이용한 Feistel 암호 알고리즘 설계

이현관 호남대학교 2002 호남대학교 학술논문집 Vol.23 No.2

This paper presents block rode algorithm which is consisted of Feistel cipher design using feedforward neural network. Using a key of the neural network's weight, it can operate the encoding and decoding. Even though it has a very complicated structure with nonlinear characterestics of neural network and huge connection within neurons, the real neurons is operating a simple process and also possible mass parallel process too. By the consisting of hidden layer, multiple design is possible.

활성화 함수의 이득 가변화를 이용한 역전파 알고리즘의 성능개선

이현관,서상혁 호남대학교 2001 호남대학교 학술논문집 Vol.22 No.2

In order to improve the several problems of the general backpropagation, we propose a method using a fuzzy logic system for automatic tuning of the activation function gain in the backpropagation. First, we researched that the changing of the gain of sigmoid function is equivalent to changing the learning rate, the weights, and the biases. The inputs of the fuzzy logic system were the sensitivity of error respect to the last layer and the mean sensitivity of error respect to the hidden layer, and the output was the gain of the sigmoid function. In order to verify the effectiveness of the proposed method, we performed simulations on the parity problem, function approximation, and pattern recognition. The results show that the proposed method has considerably improved the performance compared to the general backpropagation. Keywords: General backpropagation; Fuzzy logic system; Gain; Activation function; Parity problem; Function approximation; Pattern recognition.

신경회로망과 웨이브렛 함수를 응용한 지능제어방식

이현관 湖南大學校 情報通信硏究所 2003 정보통신연구 Vol.13 No.-

In this paper, we propose the hybrid system identification model and control method using neural networks. Proposed control method performs, for a nonlinear plant with unknown functions, system identification using an adaptive wavelet in a neural network, and then a nonlinear controller is designed with those identified informations. To identify a nonlinear function, which is not directly related to input signals, we can use an adaptive wavelet in a three-layed neural network. In order to verify the usefulness of the proposed method, we simulate the proposed control method with cart pole system.

역전파 알고리즘의 성능개선을 위한 활성화 함수의 이득 가변화 이용

이현관 호남대학교 정보통신연구소 2004 정보통신연구 Vol.14 No.-

In order to improve the several problems of the general backpropagation, we propose a method using a fuzzy logic system for automatic tuning of the activation function gain in the backpropagation. First, we researched that the changing of the gain of sigmoid function is equivalent to changing the learning rate, the weights, and the biases. The inputs of the fuzzy logic system were the sensitivity of error respect to the last layer and the mean sensitivity of error respect to the hidden layer, and the output was the gain of the sigmoid function. In order to verify the effectiveness of the proposed method, we performed simulations on the parity problem, function approximation, and pattern recognition. The results show that the proposed method has considerably improved the performance compared to the general backpropagation.

수중체중 측정법과 피하 두겹집 기법의 지방측정 비교연구

이현관 성균관대학교 체육대학 1994 체육대학 학회지 Vol.10 No.1

신경회로망을 이용한 가상 로봇 팔의 제어에 관한 연구

이현관,서상혁 호남대학교 정보통신연구소 2001 정보통신연구 Vol.11 No.-

This paper discusses an EMG based control method of a virtual robot arm as an adaptive human supporting system or remote control system, which consists of an shoulder control part, elbow control part, and wrist control part. The system uses three surface electrodes to acquire the electromyography(EMG) signal from operator. It is shown from the experiments that the EMG patterns during arm motions can be classified sufficiently by using SOFM(self-organizing feature map). The interface system based on PC environment is constructed to 3-D graphic user interface(GUI) program.

LLT 변압기 적용 LLC 직렬공진컨버터 동작특성

이현관,김은수,허동영,이기식,정봉근,강성인 전력전자학회 2006 전력전자학회 논문지 Vol.11 No.5

Operating Characteristics of LLC Series resonant converter with a LLT(Inductor-Inductor-Transformer) transformer is presented. LLT transformer used to combine the inductor and transformer into one unit has the increased leakage inductance in the primary and secondary due to the winging method and the use of the gaped core. The increased leakage inductance in the primary and secondary of LLT transformer can be impacted on the DC voltage gain characteristics of LLC series resonant converter. In the paper, DC gain characteristics and the experimental results of the LLC series resonant converter with a LLT transformer are verified on the simulation based on the theoretical analysis and the 400W experimental prototype. LLT 변압기 적용 LLC 직렬공진컨버터의 동작특성에 대해 설명하였다. 본 논문에서는 1차측 권선과 2차측 권선을 분리하고, 변압기 중간에 공극(Air-Gap)을 삽입한 일체화된 LLT 변압기를 적용하였다. 이와 같이 적용된 LLT 변압기의 경우 1차측 누설인덕턴스만이 증가되는 것이 아니고 2차측 누설인덕턴스 또한 증가되게 된다. 이렇게 증가된 2차측 누설인덕턴스는 전압이득 특성곡선에 있어 LLC 직렬공진컨버터에 적용한 변압기를 사용한 경우보다 전체적으로 높은 전압이득 특성을 가지게 된다. 따라서 본 논문에서는 2차측 누설인덕턴스를 고려한 일체화된 LLT 변압기적용 LLC 직렬공진컨버터에 대해 이론적으로 해석하고 시뮬레이션 하였으며, 400W급 시제품을 제작하여 실험한 결과에 대해 서술하였다.

네트워크 보안과 방화벽 구축 및 관리에 관한 연구

이현관 호남대학교 2003 호남대학교 학술논문집 Vol.24 No.2

An advanced tool, droplet digital PCR (ddPCR), for absolute quantification of the red-tide dinoflagellate, Cochlodinium polykrikoides Margalef (Dinophyceae)

이현관,김혜미,민주희,김근용,박명길,정해진,김광용 한국조류학회I 2017 ALGAE Vol.32 No.3

To quantify the abundance of the harmful dinoflagellate Cochlodinium polykrikoides in natural seawaters, we developed the innovative procedure using a droplet digital PCR (ddPCR) with C. polykrikoides-specific primers targeting the internal transcription sequence (ITS). The abundance of C. polykrikoides was estimated by the specific copy number of target ITS DNA segments per cell in cultures and natural water samples. The copy number per C. polykrikoides cell as acquired by ddPCR was 157 ± 16, which was evaluated against known cell numbers through a simplified protocol preparing DNAs. The abundances of C. polykrikoides in the waters of different locations estimated by ddPCR agreed with the number of cells visually counted under a microscope. This protocol was used to measure the abundance of C. polykrikoides close to and further off the southern coast of Korea in August of 2016 and 2017. The practical application showed that this method can reduce time for analysis and increase accuracy.