Prediction of fresh properties of various SCC mixes using support vector machine

Abstract

This paper presents the methodologies developed for prediction of fresh properties of Self Compacting Concrete (SCC) by using Support Vector Machine (SVM) which is an advanced statistical model. The fresh properties include L-box, slump flow, slump flow time and V-funnel. The experimental data of fresh properties of various SCC mixes has been consolidated by considering the effect of water cement ratio, water binder ratio and steel fibres. In support vector machine, Support Vector Regression (SVR) is based on the computation of a linear regression function in a high dimensional feature space where in the input data are mapped through a nonlinear function. Four models were developed to predict various fresh properties of SCC. Models were developed by using MATLAB software for training and prediction. About 75% of the data was employed for development of model and about 25% of the data is used for validation. The predicted flow properties for SCC mixes are found to be in very good agreement with those of the corresponding experimental observations available in the literature. The analysis indicates that the proposed SVM model serves as an alternative method for prediction of fresh properties of SCC.