• 시스템엔지니어링학술지
• /
• 제1권2호
• /
• pp.26-33
• /
• 2005

In this paper, an algorithm is presented to recognize road based on unpaved test courses image. The road images obtained by a video camera undergoes a pre-processing that includes filtering, gray level slicing, masking and identification of unpaved test courses. After this pre-processing, a part of image is grouped into 27 sub-windows and fed into a three-layer feed-forward neural network. The neural network is trained to indicate the road direction. The proposed algorithm has been tested with the images different from the training images, and demonstrated its efficacy for recognizing unpaved road. Based on the test results, it can be said that the algorithm successfully combines the traditional image processing and the neural network principles towards a simpler and more efficient driver warning or assistance system.

• 융합신호처리학회논문지
• /
• 제5권4호
• /
• pp.263-270
• /
• 2004

최근 웨이블릿 변환은 많은 분야에서 다양하게 적용된다. 본 논문에서 tasks뇌파의 중요한 몇가지 특성파 검출을 위한 다비치 웨이블릿은 뇌파분석에 필요하다. 우리가 제안한 시스템은 다른 방법보다는 특성파 검출에 높은 성능을 가졌다. 본 연구의 뉴럴시스템의 구조는 하나의 은닉층과 3계층 피드포워드층은 오류 BP 학습알고리즘을 적용하였다. 4명의 피험자에게 알고리즘을 적용하여 92% 분류율을 보였다. 제안된 시스템은 웨이블릿과 신경망으로 tasks 뇌파의 보다 정확하게 분석함을 보였다. 모의실험결과 tasks 뇌파는 의사의 노동력을 줄일수 있고 정량적 해석이 가능함을 보였다.

• Journal of Power Electronics
• /
• 제16권3호
• /
• pp.1176-1189
• /
• 2016

In this paper, a novel quasi-direct power control (Q-DPC) scheme based on a resonant frequency adaptive proportional-resonant (PR) current controller with a variable frequency resonant phase locked loop (RPLL) is proposed, which can achieve a fast power response with a unity power factor. It can also adapt to variations of the generator frequency in T-type Three-level shaft synchronous generator (SSG) converters. The PR controller under the static α-β frame is designed to track ac signals and to avert the strong cross coupling under the rotating d-q frame. The fundamental frequency can be precisely acquired by a RPLL from the generator terminal voltage which is distorted by harmonics. Thus, the resonant frequency of the PR controller can be confirmed exactly with optimized performance. Based on an instantaneous power balance, the load power feed-forward is added to the power command to improve the anti-disturbance performance of the dc-link. Simulations based on MATLAB/Simulink and experimental results obtained from a 75kW prototype validate the correctness and effectiveness of the proposed control scheme.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2016년도 춘계학술발표대회
• /
• pp.633-635
• /
• 2016

Convolutional Networks (ConvNets) are powerful models that learn hierarchies of visual features, which could also be used to obtain image representations for transfer learning. The basic pipeline for transfer learning is to first train a ConvNet on a large dataset (source task) and then use feed-forward units activation of the trained ConvNet as image representation for smaller datasets (target task). Our key contribution is to demonstrate superior performance of multiple ConvNet layer features over single ConvNet layer features. Combining multiple ConvNet layer features will result in more complex feature space with some features being repetitive. This requires some form of feature selection. We use AdaBoost with single stumps to implicitly select only distinct features that are useful towards classification from concatenated ConvNet features. Experimental results show that using multiple ConvNet layer activation features instead of single ConvNet layer features consistently will produce superior performance. Improvements becomes significant as we increase the distance between source task and the target task.

• 한국추진공학회:학술대회논문집
• /
• 한국추진공학회 2010년도 제35회 추계학술대회논문집
• /
• pp.499-505
• /
• 2010

다양한 비행환경에서 장시간 체공하며 운용되는 UAV에서 추진시스템을 신뢰성 있게 운용하는 것은 매우 중요하다. 이런 UAV에 사용되는 터보프롭 엔진의 정확한 손상진단은 신뢰성과 이용률을 향상시킬 수 있다. 본 연구에서는 엔진 측정 파라미터들의 변화로부터 퍼지 이론을 적용하여 손상된 구성품을 식별한 후 훈련된 신경망 알고리즘을 식별된 손상 패턴에 적용하여 손상된 양을 정확히 진단할 수 있는 방법을 제안하였다. 이렇게 제안된 진단 방법은 단일손상과 다중손상 모두 진단할 수 있다.

• Advances in concrete construction
• /
• 제3권2호
• /
• pp.91-102
• /
• 2015

In this paper, the load-deflection analysis of the Carbon Fiber Reinforced Polymer (CFRP) strengthened Reinforced Concrete (RC) slab using Recurrent Neural Network (RNN) is investigated. Six reinforced concrete slabs having dimension $1800{\times}400{\times}120mm$ with similar steel bar of 2T10 and strengthened using different length and width of CFRP were tested and compared with similar samples without CFRP. The experimental load-deflection results were normalized and then uploaded in MATLAB software. Loading, CFRP length and width were as neurons in input layer and mid-span deflection was as neuron in output layer. The network was generated using feed-forward network and a internal nonlinear condition space model to memorize the input data while training process. From 122 load-deflection data, 111 data utilized for network generation and 11 data for the network testing. The results of model on the testing stage showed that the generated RNN predicted the load-deflection analysis of the slabs in acceptable technique with a correlation of determination of 0.99. The ratio between predicted deflection by RNN and experimental output was in the range of 0.99 to 1.11.

• Structural Engineering and Mechanics
• /
• 제29권2호
• /
• pp.223-235
• /
• 2008

This paper focuses on the application of artificial neural networks (ANN) for optimal design of tensegrity grid as light-weight roof structures. A tensegrity grid, 2 m ${\times}$ 2 m in size, is fabricated by integrating four single tensegrity modules based on half-cuboctahedron configuration, using galvanised iron (GI) pipes as struts and high tensile stranded cables as tensile elements. The structure is subjected to destructive load test during which continuous monitoring of the prestress levels, key deflections and strains in the struts and the cables is carried out. The monitored structure is analyzed using finite element method (FEM) and the numerical model verified and updated with the experimental observations. The paper then explores the possibility of applying ANN based on multilayered feed forward back propagation algorithm for designing the tensegrity grid structure. The network is trained using the data generated from a finite element model of the structure validated through the physical test. After training, the network output is compared with the target and reasonable agreement is found between the two. The results demonstrate the feasibility of applying the ANNs for design of the tensegrity structures.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1998년도 제13차 학술회의논문집
• /
• pp.195-200
• /
• 1998

Since the neural network was introduced, significant progress has been made on data handling and learning algorithms. Currently, the most popular learning algorithm in neural network training is feed forward error back-propagation (FFEBP) algorithm. Aside from the success of the FFEBP algorithm, polynomial neural networks (PNN) learning has been proposed as a new learning method. The PNN learning is a self-organizing process designed to determine an appropriate set of Ivakhnenko polynomials that allow the activation of many neurons to achieve a desired state of activation that mimics a given set of sampled patterns. These neurons are interconnected in such a way that the knowledge is stored in Ivakhnenko coefficients. In this paper, the PNN model has been developed using the plasma enhanced chemical vapor deposition (PECVD) experimental data. To characterize the PECVD process using PNN, SiO$_2$films deposited under varying conditions were analyzed using fractional factorial experimental design with three center points. Parameters varied in these experiments included substrate temperature, pressure, RF power, silane flow rate and nitrous oxide flow rate. Approximately five microns of SiO$_2$were deposited on (100) silicon wafers in a Plasma-Therm 700 series PECVD system at 13.56 MHz.

• 제어로봇시스템학회논문지
• /
• 제6권1호
• /
• pp.25-34
• /
• 2000

In this paper, we design a H^\infty servo controller for gauge control of tandem cold mill. To improve the performance of the AGC(Aotomatic Gauge Control) system based on the Taylor linearized model of tandem cold mill, the H^\infty servo controller is designed to satisfy robust stability, disturbance attenuation and robust tracking properties. The H^\infty servo controller problem is modified as an usual H^\infty control problem, and the solvability condition of the H^\infty servo problem depends on the solvability of the modified H^\infty control problem. Since this modified problem does not satisfied standard assumptions for the H^\infty control problem, it is solved by an LMI(Linear Matrix Inequality) technique. Consequently, the comparison between the H^\infty servo controller and the existing PID/FF(FeedForward) controller shows the usefulness of this study.

• 제어로봇시스템학회논문지
• /
• 제12권12호
• /
• pp.1220-1230
• /
• 2006

In this paper, we propose a flew technique to the design and real-time control of an adaptive controller for robotic manipulator based on digital signal processors. The Texas Instruments DSPs(TMS320C80) chips are used in implementing real-time adaptive control algorithms to provide enhanced motion control performance for dual-arm robotic manipulators. In the proposed scheme, adaptation laws are derived from model reference adaptive control principle based on the improved Lyapunov second method. The proposed adaptive controller consists of an adaptive feed-forward and feedback controller and time-varying auxiliary controller elements. The proposed control scheme is simple in structure, fast in computation, and suitable for real-time control. Moreover, this scheme does not require any accurate dynamic modeling, nor values of manipulator parameters and payload. Performance of the proposed adaptive controller is illustrated by simulation and experimental results for a dual arm robot manipulator with eight joints. joint space and cartesian space.