Prediction of Concrete Corrosion in Sulfuric Acid by SVM-Based Method

A long time immersion test of fly ash concrete specimens in sulfuric acid is carried out. On the basis of the experimental datum, a novel prediction model for sulfuric acid corrosion of concrete based on support vector machine (SVM) is proposed. The design steps and learning algorithm are also given. Five main influencing factors including the water and cement ratio, pH value of the soaking solution, cement and fly ash quantity, specimens fluidity are analyzed with the proposed model. By comparison with the measured results, the mean square error (MSE) of predicted result of the model is 0.188. It implies that this new method of computation appears to be a useful tool for sulfuric acid corrosion of fly ash concrete research.