H.F. de Campos Velho, P. Gasbarri, L.D. Chiwiacowsky, E.H. Shiguemori (2005): A Comparison of Two Different Approaches for the Damage Idemtification Problem, 5th International Conference on Inverse Problems in Engineering: Theory and Practice (ICIPE-2005), July 11-15, Cambridge, UK.

Abstract: The damage identification problem in structural analysis can be considered as the process of determining elastic parameters through the use of numerical techniques which compare experimental measurement data. This is an important issue because identification of faults offers the prevention of future catastrophic incident and a substantial economic benefits. % The idea behind the damage identification is that the structural damage will manifest a changing in the displacement (also in velocities and accelerations) time response of the system. Structural damage detection is here displayed as an inverse vibration problem. The inverse problem, is presented as a well-posed functional form, whose solution is obtained through an optimization procedure. % The use of the artificial neural networks, a stochastic method, which has already been successfully used in thermal sciences [1], has also been presented as a satisfactory choice to deal with the damage identification problem [2]. On the other hand, among the classical methods, the use of the conjugate gradient method with the adjoint equation, or Variational Approach, has also been presented as a satisfactory choice to face the damage identification problem [3].

In this work, a comparative analysis will be presented considering the application of artificial neural network techniques and also the conjugate gradient method with the adjoint equation. These both techniques are applied to the inverse vibration problem where the goal is to estimate the unknown time-dependent stiffness coefficients simultaneously in a space truss structure. Regarding the artificial neural network, a Multilayer Perceptron Neural Network model has been employed. Numerical experiments have been carried out with synthetic experimental data considering a noise level of 1%.

References

[1] E.H. Shiguemori, H.F. de Campos Velho, J.D.S. da Silva (2004): Estimation of Initial Condition in Heat Conduction by Neural Network, Inverse Problems in Science and Engineering, 12(3) (2004), 317-328.

[2] S.V. Barai and P.C. Pandey (1997): Time-delay Neural Networks in Damage Detection of Railway Bridges, Advances in Engineering Software, 28, 1-10.

[3] L.D. Chiwiacowsky, H.F. Campos Velho and P. Gasbarri (2003): The Damage Identification Problem: A Hybrid Approach, 2nd Thematic Congress on Dynamics and Control (DINCON 2003), Sao Jose dos Campos (SP), Brazil