Institutional Repository
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/F4B3C490-8723-4147-8A2D-87AC816D0328 | ||
| Year | 2021 | ||
| Type of Item | Diploma Work | ||
| License |
|
||
| Bibliographic Citation | Georgios Gkoutzioudis, "Prediction of mechanical truss response under variable load using neural networks", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2021 https://doi.org/10.26233/heallink.tuc.88831 | ||
| Appears in Collections |
Neural networks have been applied to structural engineering in recent years. Due to the immense load of information they can analyze and process power they have and also the good results the produce they are used more and more in structural engineering. Structural analysis has also progressed a long way reaching new modernized ways making all calculations with great accuracy in modern computer systems opposing the way it was done with time consuming hand calculations. The problem with structural analysis programs used in solving design problems though is that they are often computationally expensive. Obtaining optimal solutions typically requires numerous iterations involving analysis and optimization programs. This process becomes prohibitive due to the amount of computer time required for convergence to an optimum design. Any new techniques significantly reducing the computer time required to solve design problems would be beneficial. One promising technique is to simulate a slow, expensive structural analysis program with a fast, inexpensive neural network. The purpose of this paper is to predict the mechanical response of a truss model by applying variable forces with the use of a Neural network. First of all we had to produce a large data base by analyzing a truss model with the finite element analysis (FEA) method by applying variable forces on all nodes and finding out the deformation results. We use FEA because it’s a numerical method for solving problems of engineering and mathematical physics and it is useful for problems with complicated geometries, loadings, and material properties where analytical solutions cannot be obtained like trusses.After comparing the results we obtained with other data bases and various programs, we designed and trained a ANN neural network to simulate a structural analysis program using the back-propagation algorithm Levenberg-Μarquardt to predict the mechanical response of a truss model. These procedure includes the selection of training pairs and determining the number of nodes on the hidden layer. The selection of training pairs and determining the number of nodes on the hidden layer can be found from the data bases we produced from the arithmetic simulation part of this paper.
Copyright © DIAS 2013
