• Title/Summary/Keyword: 자동미분법

Search Result 16, Processing Time 0.032 seconds

Inverse Boundary Temperature Estimation in a Two-Dimensional Cylindrical Enclosure Using Automatic Differentiation and Broyden Combined Method (자동미분법과 Broyden 혼합법을 이용한 2차원 원통형상에서의 경계온도 역추정)

  • Kim Ki-Wan;Kim Dong-Min;Baek Seung-Wook
    • Transactions of the Korean Society of Mechanical Engineers B
    • /
    • v.30 no.3 s.246
    • /
    • pp.270-277
    • /
    • 2006
  • Inverse radiation problems were solved for estimating boundary temperature distribution in a way of function estimation approach in an axisymmetric absorbing, emitting and scattering medium, given the measured radiative data. In order to reduce the computational time fur the calculation of sensitivity matrix, automatic differentiation and Broyden combined method were adopted, and their computational precision and efficiency were compared with the result obtained by finite difference approximation.. In inverse analysis, the effects of the precision of sensitivity matrix, the number of measurement points and measurement error on the estimation accuracy had been inspected using quasi-Newton method as an inverse method. Inverse solutions were validated with the result acquired by additional inverse methods of conjugate-gradient method or Levenberg-Marquardt method.

Automated algorithm of automated auditory brainstem response for neonates (신생아 청성뇌간 반응의 자동 판독 알고리즘)

  • Jung, Won-Hyuk;Hong, Hyun-Ki;Nam, Ki-Chang;Cha, Eun-Jong;Kim, Deok-Won
    • Journal of the Institute of Electronics Engineers of Korea SC
    • /
    • v.44 no.1
    • /
    • pp.100-107
    • /
    • 2007
  • AABR(automated auditory brainstem response) test is used for the screening purpose of hearing ability of neonates. In this paper, algorithm using Rolle's theorem is suggested for automatic detection of the ensemble averaged ABR waveform. The ABR waveforms were recorded from 55 normal-hearing ears of neonates at screening levels varying from 30 to 60 dBnHL. Recorded signals were analyzed by expert audiologist and by the proposed algorithm. The results showed that the proposed algorithm correctly identified latencies of the major ABR waves (III, V) with latent difference below 0.2 ms. No significant differences were found between the two methods. We also analyzed the ABR signals using derivative algorithm and compared the results with proposed algorithm. The number of detected candidate waves using the proposed algorithm was 47 % less than that of the existing one. The proposed method had lower relative errors (0.01 % error at 60dBnHL) compared to the existing one. By using proposed algorithm, clinicians can detect and label waves III and V more objectively and quantitatively than the manual detection method.

Stress Constraint Topology Optimization using Backpropagation Method in Design Sensitivity Analysis (설계민감도 해석에서 역전파 방법을 사용한 응력제한조건 위상최적설계)

  • Min-Geun, Kim;Seok-Chan, Kim;Jaeseung, Kim;Jai-Kyung, Lee;Geun-Ho, Lee
    • Journal of the Computational Structural Engineering Institute of Korea
    • /
    • v.35 no.6
    • /
    • pp.367-374
    • /
    • 2022
  • This papter presents the use of the automatic differential method based on the backpropagation method to obtain the design sensitivity and its application to topology optimization considering the stress constraints. Solving topology optimization problems with stress constraints is difficult owing to singularities, the local nature of stress constraints, and nonlinearity with respect to design variables. To solve the singularity problem, the stress relaxation technique is used, and p-norm for stress constraints is applied instead of local stresses for global stress measures. To overcome the nonlinearity of the design variables in stress constraint problems, it is important to analytically obtain the exact design sensitivity. In conventional topology optimization, design sensitivity is obtained efficiently and accurately using the adjoint variable method; however, obtaining the design sensitivity analytically and additionally solving the adjoint equation is difficult. To address this problem, the design sensitivity is obtained using a backpropagation technique that is used to determine optimal weights and biases in the artificial neural network, and it is applied to the topology optimization with the stress constraints. The backpropagation technique is used in automatic differentiation and can simplify the calculation of the design sensitivity for the objectives or constraint functions without complicated analytical derivations. In addition, the backpropagation process is more computationally efficient than solving adjoint equations in sensitivity calculations.

패턴 인식기법을 이용한 유출모형의 매개변수 최적화

  • 정창삼;허준행
    • Proceedings of the Korea Water Resources Association Conference
    • /
    • 2002.05b
    • /
    • pp.1316-1321
    • /
    • 2002
  • 일반적으로 강우-유출모형은 lumped model과 distributed model로 크게 구분될 수 있으며, 우리나라에서는 이중 비교적 부족한 자료를 이용하여도 개략적 모의가 가능한 전자를 널리 사용하고 있다. 본 연구에서는 이러한 모형들의 매개변수를 보정하는 방법에 관해 연구하였다. 일반적으로 모형의 보정 방법에는 크게 시행오차에 의한 수동보정(manual calibration) 방법과 최적화 기법에 의한 자동보정(automatic calibration) 방법으로 나눌 수 있다. 수동보정 방법은 모형 수행결과를 수문곡선의 시각적 비교에 의해 관측치와 비교하여 모형 운영자의 주관적인 판단하에 조정하는 기법이며, 자동보정 방법은 최적화 기법을 이용8하여 특정한 산정기준(estimation criteria)을 최대 또는 최소화시켜 모형의 매개변수를 결정하는 방법이다. 이러한 최적화기법은 일반적으로 직접탐색법과 경사법으로 구분할 수 있다. 경사법은 수렴속도가 빠르지만 편미분에 의해 방향을 찾아가는 방법으로 편도함수가 필요하므로 수문모형에는 적용하기가 힘들므로 적합하지 않다. 그러나, 보다 많은 컴퓨터 수행시간을 필요로 하는 직접탐색법의 경우 수렴속도는 느리지만, 편도함수를 필요치 않으므로 수문모형의 최적화 기법으로 적합하다고 할 수 있다. 직접탐색법에는 simplex-search 법, 패턴인식(pattern-search)법, rotating-direction 법, brent 법 등이 있으며, 본 연구에서는 직접탐색법의 일종인 패턴인식(pattern -search)법을 이용하여 매개변수 최적화 과정을 모의하였다. 이러한 매개변수 보정모형을 구성한 후 이를 가장 보편적으로 사용되고 있는 유출모형인 각종 단위도법들을 결합하는 모형을 구성하였다. 또한 구성된 모형을 시범유역에 적용하여 나온 결과를 HEC-1에서 적용되고 있는 단일변량 증감법과 같은 최적화 기법을 이용한 결과와 비교·분석을 실시하였다. 본 모형을 활용하여 강우-유출 모형의 매개변수를 지속적으로 산정하고 일반화할 경우 임의의 유역의 수문기상학적 특성에 부합한 매개변수를 정량화 시킬 수 있었다.

  • PDF

On the Implementation of PDE Solver using Finite Element Method in CEMTool (CEMTool에서 유한요소법을 이용한 PDE Solver구현에 관하여)

  • Ahn, Choon-Ki;Lee, Tae-Il;Han, Soo-Hee;Lee, Young-Sam;Kwon, Wook-Hyun
    • Proceedings of the KIEE Conference
    • /
    • 2003.07d
    • /
    • pp.2594-2596
    • /
    • 2003
  • 유한요소법(FEM)은 많은 공학문제를 해결하는 가장 중요한 방법 중 하나로 인식되고 있다. 본 논문에서는 자동제어 및 신호처리 문제해결에 효율적이며 강력한 수치해석 패키지인 CEMTool환경에서 유한요소법을 이용하여 일반적인 편미분방정식 Solver 구현에 관한사항을 논의하고자 한다. 기본적으로 영역정보 및 노드수 등의 정보를 입력받아 각 노드의 정보를 출력하는 Mesh함수를 구현하며, 생성된 요소방정식들을 조립하는 Assemble함수를 작성한 뒤, Boundary함수를 통해 경계조건을 적용시킨 후 선형행렬 방정식을 풀어 전체노드의 값을 찾아내는 Solve함수를 구현하는 과정을 알아본다. 구현된 FEM Solver의 전체적인 구조를 통해 구현시 고려해야 할 사항을 논의하며 기본적인 편미분방정식의 예제를 통해 FEM PDE Solver의 동작과정을 검증할 것이다.

  • PDF

An Improved Multi-level Optimization Algorithm for Orthotropic Steel Deck Bridges (강바닥판교의 개선된 다단계 최적설계 알고리즘)

  • 조효남;이광민;최영민;김정호
    • Journal of the Computational Structural Engineering Institute of Korea
    • /
    • v.16 no.3
    • /
    • pp.237-250
    • /
    • 2003
  • Since an orthotropic steel deck bridge has large number of design variables and shows complex structural behavior, it would be very difficult and impractical to directly use a Conventional Single Level (CSL) optimization algorithm for its optimum design. Thus, in this paper, an Improved Multi Level Design Synthesis (IMLDS) optimization algorithm is proposed to improve the computational efficiency. In the proposed IMLDS algorithm, a coordination method is introduced to divide the bridge into main girders and orthotropic steel deck with preserving the characteristics of the structural behavior. For an efficient optimization of the bridge, the IMLDS algorithm incorporates the various crucial approximation techniques such as constraints deletion, Automatic Differentiation (AD), stress reanalysis, and etc. In the case of orthotropic steel deck system, optimum design problems are characterized by mixed continuous discrete variables and discontinuous design space. Thus, a modified Genetic Algorithm (GA) is also applied to optimize discrete member design for orthotropic steel deck. From the numerical example, the efficiency and convergency of the IMLDS algorithm proposed in this paper is investigated. It may be positively stated that the IMLDS algorithm will lead to more effective and practical design compared with previous algorithms.

A Study on Automatic Seam Tracking System Using Electro-Magnetic Sensor for Sheet Metal Arc Welding of Butt Joints (박판 맞대기 용접에서 전자기식 센서를 이용한 용접선 자동 추적 시스템에 관한 연구)

  • 유병희;김재웅
    • Journal of Welding and Joining
    • /
    • v.15 no.1
    • /
    • pp.81-91
    • /
    • 1997
  • In this study, a magnetic sensor to make use of eddy current was developed to detect the weld seam of butt joint in the sheet metal arc welding. This system consist of the sensor device for detecting the weld line, the servo control device for driving the weld torch movement and the control unit. A signal processing was applied to smooth the output signal of the sensor. The weld joint was determined by using a 1st order differential method. To improve tracking accuracy of the system, moving average method which has an effect of proportional and weighted integral control was applied to a series of the weld joint positions obtained above. The weld line for tracking was generated by using data regeneration algorithm. Based on these results, each servo motor was controlled by pulse generator. From experimental results, it was revealed that this system has excellent detecting ability for weld line and seam tracking ability.

  • PDF

Controller Auto-tuning Scheme using System Monitoring inFrequency Domain (주파수역에서 시스템 감시를 이용한 제어기 Auto-tuning기법)

  • 정유철;이건복
    • Proceedings of the Korean Society of Precision Engineering Conference
    • /
    • 2000.11a
    • /
    • pp.136-139
    • /
    • 2000
  • Controller refinement scheme to improve the performance of a conventional system automatically in frequency domain is proposed. The controller automatic tuning method features using experimental frequency responses of the conventional closed-loop system, the conventional controller, and the improved closed-Imp system; instead of poorly modeled plant due to non-linearities and disturbances. The improved closed-loop system characteristics is automatically acquired by the conventional closed-loop system characteristics and the proposed performance index in system bandwidth. And the proper controller is realized by least squares approximation in frequency domain. To testify the usefulness of the approach, experimental results of robot path-tracking control applied with various controllers is used, and then is analyzed with respect to a equivalent proportional controller. Experimental results and analytic results are well-matched.

  • PDF

A Study on the Pattern Segmentation and Classification in Specially Documentated Imaged (제한된 문서 영상에서 패턴 분절과 구분 처리에 관한 연구)

  • 옥철호;허도근;진용옥
    • The Journal of Korean Institute of Communications and Information Sciences
    • /
    • v.14 no.6
    • /
    • pp.663-674
    • /
    • 1989
  • In order to design the automatic processing system of image document, the pattern segmentation of image document and classification methods are presented. The contour extraction using first order differential operator of Gauassian distribution fucntions, the image segmentation using the chain code, and the pattern classication using the second order moments and two=dimensional Rf distance(in transform domain) are implemented. The resuts applied in specially documantated image shows to classify the characters, fingerprints, seals etc well. And the utility of the used algorithms is verified.

  • PDF

Nonlinear PID Controller with Simple Neural Network Structure (간단한 신경회로망 구조를 갖는 비선형 PID 제어기)

  • 정경권;김주웅;정성부;김한웅;엄기환
    • Proceedings of the Korean Institute of Information and Commucation Sciences Conference
    • /
    • 1998.05a
    • /
    • pp.96-101
    • /
    • 1998
  • 많은 분야에서 널리 사용되고 있는 PID 제어기의 형태는 오차를 갖는 폐루프 시스템으로 구성되며, PID 제어기는 비례, 적분, 미분 제어기로 나누어진다. PID 제어기의 형태가 여러 가지로 제안되고 있지만 보다 중요한 것은 PID 제어기의 파라미터들을 어떻게 적절히 정하느냐 하는 파라미터 조정 문제이다. 실제로 산업 현장에 설치되어 있는 PID 제어기는 대부분 숙련된 기술자에 의해 수동 조작에 의한 시행 착오(trial and error) 법으로 동조되고 있다. 이 경우는 많은 노력과 시간이 소비되고, 외란(disturbance)이 첨가될 경우 적절히 동조된다는 보장도 없다. 본 논문에서는 이러한 문제를 해결하고자 신경회로망을 이용하여 PID 제어기의 파라미터를 동조하는 제어 방법을 제안하였다. 단일 뉴런으로 구성하여 구조가 간단하고, 학습에 의한 성능 개선이 가능하다. 오차 역전파(Error Back-Propagation) 알고리즘에 의하여 PID 파라미터가 되는 가중치를 자동 동조하는 방법이다. 제안한 방식의 유용성을 보이기 위해 DC 서보 모터와 비선형 시스템인 단일 관절 매니퓰레이터를 대상으로 시뮬레이션을 하였다.

  • PDF