Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 97
PROCEEDINGS OF THE SECOND INTERNATIONAL CONFERENCE ON SOFT COMPUTING TECHNOLOGY IN CIVIL, STRUCTURAL AND ENVIRONMENTAL ENGINEERING
Edited by: Y. Tsompanakis, B.H.V. Topping
Paper 31

Analysis of Retractable Roof Structures

S. Kmet and J. Cholvadt

Faculty of Civil Engineering, Technical University of Kosice, Slovakia

Full Bibliographic Reference for this paper
S. Kmet, J. Cholvadt, "Analysis of Retractable Roof Structures", in Y. Tsompanakis, B.H.V. Topping, (Editors), "Proceedings of the Second International Conference on Soft Computing Technology in Civil, Structural and Environmental Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 31, 2011. doi:10.4203/ccp.97.31
Keywords: retractable roof structure, finite element analysis, geometrically nonlinear analysis, computational models, parametric study, artificial neural networks, learning techniques, activation functions, multilayer perceptron, structural behaviour prediction, rail track displacements.

Summary
Large span retractable roof structures, compared to traditional fixed systems have many specifics [1]. These relate mainly to the movement of the mobile segment and using the structure in different states.

In this paper an application of artificial neural networks (ANNs) for the analysis and behaviour prediction of the selected large-scale retractable roof structure under varying load conditions and states of mobile roof segments (an open, semi-open and closed state) is presented. At first, the geometrically nonlinear response of the structure in the form of displacements of the rail track for the two movable segments located on the two parallel spatial truss arches with a span of 224m is analyzed using the finite element method (FEM). As a consequence, ANNs are trained and tested to replace the finite element (FE) analysis and to predict the required structural response characteristics for the loading intervals and different states of mobile roof segments considered.

The SCIA Ida Nexis 32 software [2] was used for the analysis of the structure to acquire the necessary quantity of relevant data for the training and testing of ANNs. Transverse and longitudinal forces arising from the movement of the movable roof and a wind load affecting the movable segments located at several different positions were the main four variables considered in this study. Totally, 360 different load combinations were generated. The displacements are strongly nonlinear and that is why the use of ANNs is suitable.

Numerical experiments were made with generalized residual neural networks, radial basis functions networks and multilayer perceptron neural networks. The influence of different network topologies, learning algorithms, synaptic and activation functions was tested. All experiments were realized by the StatSoft Statistica 7 software with the neural network superstructure [3]. Totally, 360 files obtained from the nonlinear FE analysis were randomly divided into the training, validation and testing sets in the ratio 2:1:1, respectively.

The best results were reached by the perceptron neural network with the topology 4-79-42-42 and backpropagation learning algorithm in the combination with the conjugate gradient algorithm. For this topology the mean square error MSE = 3.3% during the training procedure and MSE = 3.5% in the testing phase were achieved.

The results confirmed that ANNs can be applied for a behaviour prediction and analysis of retractable roof structures under current load conditions in real time and can be effectively used in their control systems.

References
1
K. Ishii, "Structural design of retractable roof structures", WIT Press, Southampton, United Kingdom, 2000.
2
Ida Nexis 32, "User manual", SCIA, 1998.
3
Statistica 7 Neural Networks, "Electronic manual", StatSoft, 2004.

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