Modelling Initial Design of Steel Portal Frame using ANN

Autor/innen

  • Vasudev R. Upadhye Department of Civil Engineering, Marathwada Institute of Technology, Aurangabad (M. S.), India
  • Chandrashekhar P. Jagtap Department of Civil Engineering, Marathwada Institute of Technology, Aurangabad (M. S.), India
  • Ajay G. Dahake Department of Civil Engineering, Marathwada Institute of Technology, Aurangabad (M. S.), India

Abstract

Structural Engineering involves the understanding and modeling of natural phenomenon, material behavior and laws of mechanics, intuition, past experience or expertise and knowledge of analysis techniques. Especially in structural design, the difficulty in problem definition and the large number of constraints makes it difficult to program a machine for computerization of the total design process. The analysis of structure is performed based on the initial design and redesign of the structure is done using the results of this analysis.  Various softwares available based on stiffness matrix method or finite element method is also required to be provided with the geometric properties of the sections initially. To reduce number of subsequent cycles for analysis and design a good initial design model is required as it is very difficult to interpret the correct cross section and reinforcement etc. Artificial neural network (ANN) is a correct tool as it can be learnt from available designs. It is observed from the literature review that ANN can be used effectively in structural optimization, initializing the design procedure of structural elements. In this research, an attempt is made to apply ANN for modeling the initial design of portal rigid steel frame. A dataset over certain range is developed for training and testing of ANNs. The results of the unseen problems by the network are compared with the dataset and prediction of initial design of frame by the network.  Keywords: ANN, beam, column, steel, portalCite this Article Upadhye et al. Modelling Initial Design of Steel Portal Frame using ANN. Journal of Aerospace Engineering and Technology. 2016; 6(2): 16–20p.

Veröffentlicht

2016-07-04

Ausgabe

Rubrik

Research Articles