Loading...

Development and Assessment of Meshfree Neural Network Based Methods for Linear and Nonlinear Static Analysis of Structures

Ara, Kiarash | 2013

458 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 45745 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Joghataie, Abdolreza
  7. Abstract:
  8. Modeling, static and dynamic analyses of structures are main subject of research and application in structural engineering. Also innovative methods gradually substitute for conventional methods in any field of science and technology. Nowadays neural networks have attracted special attention in any branch of science and engineering. Some structural engineering researchers have utilized neural networks as expert systems in field of material modeling, static and dynamic analysis, active control and etc. The neuro-finite element method is a neural network based method for linear and nonlinear static and dynamic analysis of structures with one-dimensional members. In this method neural networks have been trained based on combining elements behavior. The goal of this thesis is to develop neuro-finite element method for analyzing the structures with continuous domain, like bar elements, beams and plate and introducing meshless neuro-finite element method. In this approach, an algorithm is presented for selecting the most appropriate shape functions for meshless finite element analysis using neural networks. The training data is provided by nonlinear numerical analysis of the structure under study for different loading cases corresponding to a variety of shape functions. Hence, the algorithm is called Meshless Neuro-Finite Element algorithm. Multi-layer feedforward neural networks and Prandtl neural networks are used
  9. Keywords:
  10. Nonlinear Static Analysis ; Neural Networks ; Neuro-Finite Elements Method ; Meshless Method

 Digital Object List

 Bookmark

No TOC