Prediction of Compressive Strength for different Curing Stages using Steepest Descent ANNs

Conference: International conference on Sustainable Design, Engineering and Construction
Author(s): K. Rajasekhar, Gottapu Santosh Kumar Year: 2016
Grenze ID: 02.SDEC.2016.1.507 Page: 62-65

Abstract

A new ANNs training model is proposed to predict the compressive strength of Slurry Infiltrated Fibrous Concrete (SIFCON) with manufactured sand for various curing periods in this paper. All over the world, availability of natural sand is dwindling day by day due to its limited availability in the river beds. The alternative to this natural resource is the manufactured sand. The ANNs model is constructed, trained and tested for predicting the compressive strengths of various ages using Steepest Descent based weights updating method. The experimental data corresponding to compressive strength of the SIFCON specimens are obtained using the Universal Testing Machine (UTM) of capacity 300T. A total of 80 different SIFCON cube specimens of M20 grade mix using the manufactured sand is cast for different fibre fraction (8%, 10% and 12%) and tested for a different curing period such as 7, 28 and 56 days. In this paper, the compressive strength of different curing ages (14, 21, 35, 42, 49, 63, 70, 77, 84 and 91 days) is predicted for 8%, 10% and 12% fibre volume fraction. For the ANN training phase, different fibre configurations, curing of SIFCON cubes corresponding to various time periods, number of neurons, learning rate, momentum and activation functions were considered. The results show that the relative percentage error in the training set was 6.02% and the testing set was 15%. Research results demonstrate that the proposed Steepest Descent based ANN model is practical, gives high prediction accuracy and beneficial.

<< BACK

SDEC - 2016