Strength prediction using ANN for concrete with Marble and Quarry dust

Abstract

Modern construction material research is picking impetus in the recent two decades; a greater number of admixtures and combinations were tried by bountiful researchers across the globe. In this work an attempt is made to obtain the strength characteristics by using Soft computing techniques in the marble and quarry dust impregnated concrete. Strength characteristics of concrete is studied with reference to the addition of the above-mentioned admixtures and the results were given as input parameters. 28 days compressive strength of concrete with varying marble and quarry dust content is utilized as input data for the neural network and a model is created which is used to predict the strength. To prepare the ANN model the results are taken and the values obtained are mean square propagation and the testing, training, validation and for overall propagation the values are 0.99793, 0.99577, 0.9927 and 0.99073 and the best validation performance is 0.023295 at epoch 7 respectively for MD and for QD the values are 0.9974, 0.94374, 0.94445 and 0.947 and the best validation performance is 0.035578 at epoch 4 respectively. It is found that neural network can be utilized effectively to predict the strength characteristics of concrete.