A robust neural network classifier to model the compressive strength of high performance concrete using feature subset selection

  • Authors:

  • M. Venu;R. Uday Kiran;R. Kiranmai

  • Affiliations:

  • Birla Institute of Technology and Science, Pilani, Andhra pradesh, India;International Institute of Information Technology, Hyderabad, Andhra pradesh, India;Indur Institute of Engineering and Technology, Medak, Andhra pradesh, India

  • Venue:
  • Proceedings of the 5th ACM COMPUTE Conference: Intelligent & scalable system technologies
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

High performance concrete (HPC) is a mixture of cement, fine aggregate, coarse aggregate, water and other ingredients. Modeling the compressive strength of HPC (or concrete strength) is a difficult task in building materials because it is influenced by the proportions of various ingredients within the HPC. Researchers have tried to confront this difficulty by modeling the strength of concrete using artificial neural networks (ANN). The influence of ingredients on concrete strength still remains unknown due to the "black box" nature of ANN. This paper investigates the influence of ingredients on concrete strength modeling using feature selection. A robust ANN-model using the knowledge from feature selection has been developed to predict the strength effectively. Experimental results have shown that the proposed model is efficient with respect to runtime and prediction.