|
NISCAIR ONLINE PERIODICALS REPOSITORY (NOPR) >
NISCAIR PUBLICATIONS >
Research Journals >
Indian Journal of Engineering and Materials Sciences (IJEMS) >
IJEMS Vol.11 [2004] >
IJEMS Vol.11(1) [February 2004] >
| Title: | Backcalculation of pavement layer parameters using Artificial Neural Networks |
| Authors: | Saltan, Mehmet
Terzi, Serdal |
| Issue Date: | Feb-2004 |
| Publisher: | CSIR |
| Abstract: | In this paper, a new formulation
based on Artificial Neural Networks (ANN) is presented for backcalculation of
pavement layer moduli. In structural analysis of flexible pavements, the
procedures as Layered Elastic Theory, Equivalent Layer Thickness (ELT), and
Finite Elements Method (FEM) generally have complex formulations and give
approximate results. Therefore, it is extremely difficult to perform realistic
analysis for flexible pavements, especially in view of modelling the material
properties of layers in these methods. Setting the finite element mesh and
iteration procedure of backcalculation takes rather long time. The proposed ANN
procedure requires significantly less computation time. ELT method is used for
simplicity. It is impossible or very hard to model the visco-elastic and
non-linear behaviour of layer materials in layered elastic theory. The use of
ANN is proliferating with high rate in simulation. The ability of ANN is to
learn complex nonlinear relationships. A new formulation using ANN is presented
here. |
| Page(s): | 38-42 |
| ISSN: | 0975-1017 (Online); 0971-4588 (Print) |
| Source: | IJEMS Vol.11(1) [February 2004]
|
|
