Prediction of Elastic Modulus of Concrete Using Support Vector Committee Method

Abstract

Knowledge about concrete properties is of utmost importance in engineering materials, and elastic modulus is one of concrete’s most important properties that is used in the calculation of deformation of structures. For this reason, many researchers have attempted to introduce various correlations between this property and the compressive strength. In this paper, support vector committee (SVC) is used for prediction of elastic modulus of normal strength (NSC) and high-strength concrete (HSC). The SVC is based on learning theory, and deploys the technique by introducing accuracy insensitive loss function. The comparison between concrete elastic modulus predicted by the SVC method with the experimental data and those from other methods like support vector machine (SVM), artificial neural networks (ANN), fuzzy logic, and other conventional methods show marked improvement in relation to the best of prediction methods with error indices constantly less than 1%. It is therefore concluded that the SVC model is a greatly more effective method of prediction for elastic modulus of all grades of concrete.