An optimised support vector machine model for elastic modulus prediction of concrete subject to alkali silica reaction

Alkali-silica reaction (ASR) can induce the damage and loss in serviceability of concrete structures. Many studies have been conducted to investigate the influence of ASR on the degradation of mechanical properties of the concrete. Their results show that compared with other mechanical properties, the modulus of elasticity is the most affected by ASR, where the reduction is up to roughly 70% compared to its properties without expansion. In this study, to effectively assess the reduction of the modulus of elasticity caused by ASR, a novel predictive model is proposed based on support vector machine (SVM), in which the mix proportion of concrete, exposure environment and corresponding expansion are employed as the inputs and the output is the modulus of elasticity degradation. To improve the generalization capacity of the proposed predictive model, three different optimization algorithms are adopted to select optimal model parameters. Finally, the experimental data from the existing literatures are used to test the performance of the proposed method with satisfactory results