한국지진공학회논문집 (Journal of the Earthquake Engineering Society of Korea)

  • 제2권1호
  • /
  • Pages.35-46
  • /
  • 1998
  • /
  • 1226-525X(pISSN)
  • /
  • 2234-1099(eISSN)
한국지진공학회 (Earthquake Engineering Society of Korea)
권호 표지

신경망을 이용한 구조물 접합부의 손상도 추정

Structural Joint Damage Assessment Using Neural Networks

  • 방은영 (Univ. of California, Irvine, 박사후연구원과정) ;
  • 이진학 (한국과학기술원 토목공학과) ;
  • 윤정방 (한국과학기술원 토목공학과)
  • 발행 : 1998.03.01
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

대부분의 손상도 추정법들을 부재의 손상을 해당부재의 평균적인 강성감소로 표현하였다. 본 연구에서는 보다 실제적인 손상도를 추정하기 위하여, 접합부의 손상을 도입하였다. 접합부의 모형화를 위하여 보의 양단에 회전스프링을 추가하였으며, 접합부 손상을 접합부 강성의 감소로 정의하였다. 접합부의 손상도를 계측된 모드벡터를 바탕으로하여, 신경망기법을 추정하였다. 효율적인 훈련패턴을 만들기 위하여 Latin Hypercube Sampling 기법을 도입하였으며, 국부영역에서의 손상도추정을 위하여 부구조법을 도입하였다. 제안된 기법의 효율성을 검증하기 위하여 10층 프레임구조물에 대한 수치해석결과를 이용하였다. 예제해석을 통하여 추정결과가 상당히 정확함을 확인하여, 실제 적용 가능한 방법임을 알수 있었다.

Structural damage is used to be modeled through reductions in the stiffness of structural elements for the purpose of damage estimation of structural system. In this study, the concept of joint damage is employed for more realistic damage assessment of a steel structure. The joint damage is estimated damage based on the mode shape informations using neural networks, The beam-to-column connection in a steel frame structure is represented by a rotational spring at the fixed end of a beam element. The severity of joint damage is defined as the reduction ratio of the connection stiffness with respect to the value of the intact joint. The concept of the substructural identification is used for the localized damage assessment in a large structure. The feasibility of the proposed method is examined using an example with simulated data. It has been found that the joint damages can be reasonably estimated for the case with the measurements of the mode vectors subjected to noise.

키워드

참고문헌

  1. J. Engrg. Mech. v.121 no.2 Structural-system identification I: Theory Chanem, R.;Shinozuka, M.
  2. Comput. & Struct. v.26 no.5 Effects of joint flexibility on the behavior of steel frames Chen, W. F.;Lui, E. M.
  3. J. Struct. Engrg. v.120 no.4 Nonlinear vibration analysis of steel frames with semirigid connections Chan, S. L.;Ho, G.W.M.
  4. J. of Structral Engrg v.122 no.11 Modelling for momentrotation characteristics for end-plate connections Shi, Y.J.;Chan, S.L.;Wong, Y.L.
  5. Can. J. Civil Eng. v.2 Analysis of flexibly connected steel frames Frye, J. M.;Morris, G. A.
  6. J. Struct. Engrg. v.116 no.7 Moment-rotation relations of semirigid connections with angles Kishi, N.;Chen, W. F.
  7. Comput & Struct. v.42 no.4 Use of neural networks in detection of structural damage Wu, X.;Ghaboussi, J.;Garrett J.H.Jr
  8. Earthquake Engrg. and Struct. Dynamics v.20 Estimation of structural parameters in the time domain: a substructure approach Koh, C. G.;See, L. M.;Balendra, T.
  9. Numerical Recipes in C-The Art of Scientific Computing Press. W.;Teukolky, S.A.;Vetterling, W.T.Flannery, B. P.
  10. AIAA J. v.6 no.7 Compling of substructures for dynamic analysis Craig, R.R.Jr.;Bampton, M.C.C.
  11. Theory of Probability and Its Applications v.16 no.2 On the uniform convergence of relative frequencies of events to their probabilities Vapnik, V. N.;Chervonenkis, A. Ya.
  12. IEEE Trans. Systems, Man, and Cybernetics v.22 no.3 Noise injection into inputs in back-propagation learning Matsuoka, K.