Keywords: neural network, modelling, identification, building, inverse vibration, multistorey.

Even for the systems, which may be modelled as linear, the identification problem often turns in to a non-linear optimization problem. This requires an intelligent iterative scheme to have the required solution. There are various on-line and off-line methods, viz. the Gauss-Newton, Kalman filtering and probabilistic methods such as maximum likelihood estimation etc. However the identification problem for a large number of parameters, the following two basic difficulties are faced often:

1. The objective function surface may have multiple maxima and minima and the convergence to the correct parameters is possible only if the initial guess is considered as close to the parameters to be identified; and
2. The inverse problem in general gives non-unique parameter estimates.

To overcome these difficulties, the present study developed an identification methodology for the said problem by the use of powerful technique of Artificial Neural Network (ANN). However, recently a number of studies: Masri et al. [1], Chassiakos and Masri [2], Narendra and Parthasarathy [3], Bani-Han et al. [4], Huang et al. [5], Chakraverty [6] and the references mentioned there in used ANNs for the structural identification problems. It is to be noted that these studies (using ANNs) identify in general the restoring force of the system using input/output data.

This paper demonstrates the use of powerful technique of Artificial Neural Network (ANN) for the identification of structural parameters of multistorey shear building using the response of the structure subject to horizontal (ground) displacement. Here, for given input to the system, rather than solving the inverse vibration problem, the forward problem for each time step has been solved as usual to generate the solution vector. First the initial (prior) values of the physical parameters (stiffness etc.) of the system are randomized for the numerical experiment and then using these sets of physical parameters, the responses have been obtained. The responses and the corresponding parameters are used as the input/output in the neural net. An iterative scheme is proposed to train the neural network. When the iterative training of the network is done for an acceptable accuracy the final converged weight matrix is obtained. Then the physical parameters may be identified if new response data is supplied as input to the net. The procedure has been demonstrated for multistorey structure and the structural parameters are identified using the response of the structure subject to horizontal (ground) displacement. The model has been tested for the identification of the stiffness parameters of multistorey structure using the prior values of the design parameters and the results are found to be reliable and comparable.

The value of the paper concerns mainly the use of ANN iteratively in particular for system identification problems for the first time and in obtaining the weight matrices similar to memory matrices. Moreover only the design parameters have been used for the training of the network. The developed methodology may be used to have the trained neural model which may identify the parameters of the structure by utilizing only the corresponding final converged weights for the system in the future.

1

Masri, S.F., Smyth, A.W., Chassiakos, A.G., Caughey, T.K., and Hunter, N.F. "Application of neural networks for detection of changes in nonlinear systems." J. of Engineering Mechanics (ASCE), 126(7), 666-676, 2000 . doi:10.1061/(ASCE)0733-9399(2000)126:7(666)

2

Chassiakos, A.G., and Masri, S.F. "Modelling unknown structural systems through the use of neural networks." J. Earthquake Engng. And Struct. Dyn., 25, 117-128, 1996. doi:10.1002/(SICI)1096-9845(199602)25:2<117::AID-EQE541>3.0.CO;2-A

3

Narendra, K.S., and Parthasarathy, K. "Identification and control of dynamical systems using neural networks." IEEE Trans. Neural Networks, 1, 4-27, 1990. doi:10.1109/72.80202

4

Bani-Han, K., Ghaboussi, J. and Schneider, S.P. "Experimental study of identification and control of structures using neural network : Part 1 : Identification." J. Earthquake Engng. And Struct. Dyn., 28, 995- 1018, 1999. doi:10.1002/(SICI)1096-9845(199909)28:9<995::AID-EQE851>3.0.CO;2-8

5

Huang, C.S., Hung, S.L., Wen, C.M., and Tu, T.T. "A neural network approach for structural identification and diagnosis of a building from seismic response data." J. Earthquake Engng. And Struct. Dyn., 32, 187-206, 2003. doi:10.1002/eqe.219

6

Chakraverty, S., Sharma, R.K., and Singh, V.P. "Soft-computing approach for identification of dynamic systems." J. New Build. Mat. & Const. World, 9(2), 50-56, 2003.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £95 +P&P)