Checking the Homogeneity of Concrete Using Artificial Neural Network
Tomme: LXI (LXV) Fascicle: 3 | 2015
Pages: 83-92
Abstract text:
This paper proposes a manner to verify the concrete samples homogeneity using artificial neural networks. This method determines the percentages of different areas of component materials visible at top and bottom of a concrete cylinder having 20 cm diameter and 20 cm height. The materials that have been achieved are rubber grains, aggregates and mineral matrix. The training of the neural network was realise by using backpropagation algorithm and then, in order to separate the regions of interest was used Levenberg – Marquardt algorithm. As a neural network input data were used photos having 258 ×170 pixels resolution achieved both on the both sides of the cylinder.
Key Words:
neural network; backpropagation algorithm; concrete; rubber; computing percentage; Levenberg – Marquardt algorithm
Author(s) Information
All documents with a 
icon require
Adobe Acrobat installed on your computer
