Neural Models for Unconfined Compressive Strength of Kaolin clay mixed with pond ash, rice husk ash and cement

Abstract

In this study an Artificial Neural Network (ANN) model was used to predict the Unconfined Compressive Strength (UCS) of Kaolin clay mixed with pond ash, rice husk ash and cement content model under different curing period. The input parameters included percentages of admixtures added alongwith clay content and curing period. The curing Period range was 7, 14 and 28 days considered in neural model. The feedforward backpropagated neural model with Levenberg Marqaurdt gradient descent with momentum constant was used to predict the UCS and optimized topology of 5-10-1 was obtained. The sensitivity analysis based on weights of neural model indicated that all admixtures contributed 70% to the UCS of Kaolin clay. The comparison of ANN model with Multiple Regression Analysis (MRA) model indicated that ANN models were performing better than MRA model with values of r as R2 as 0.98 and 0.97 respectively in testing phase of neural model and for MRA model r was 0.94 and R2 as 0.88.