Scientific Research and Essays

  • Abbreviation: Sci. Res. Essays
  • Language: English
  • ISSN: 1992-2248
  • DOI: 10.5897/SRE
  • Start Year: 2006
  • Published Articles: 2754

Full Length Research Paper

Application of Neural Network models on analysis of prismatic structures

  N. TayÅŸi    
Department of Civil Engineering, Faculty of Engineering, University of Gaziantep,   27310 Gaziantep, Turkey.
Email: [email protected]

  •  Accepted: 06 April 2010
  •  Published: 31 May 2010



A Neural Network based design system is presented in this paper for preliminary analysis and design of prismatic structures. Because of the broad diversity of prismatic structures encountered in practice, it becomes clear that this study is concentrated on fundamental frequencies of single cell box girder bridge with straight planform. To provide a wide range of dataset for neural network training, fundamental frequencies are determined using variable thickness Mindlin-Reissner finite strips which offers an accurate and inexpensive tool for the production of database. The results of proposed neural network model to compute the fundamental frequency are furthermore compared with the results of finite strip analysis and found to be quite accurate. The trained neural network model proposed in this study and finite strip methods are used to conduct an extensive parametric study to investigate its generalization capability and the effect of various parameters on the fundamental frequency.


Key words: Soft computing, Finite strip method, free vibration, Neural Networks.