Artificial Intelligence for Compressive Strength Prediction of Concrete

Abstract

Structural health monitoring is an indispensable procedure that is to be carried out to evaluate the serviceability of existing structures. Non-destructive testing methods are attaining increasing popularity for the assessment of concrete strength due to the ease of operation and reliability of the results. The application of machine learning in the field of engineering has increased rampantly. In this study, the compressive strength of concrete has been forecasted using support vector regression which is a machine learning technique. Artificial intelligence is nothing but the potential to impersonate human intelligence. The advantage of artificial intelligence over human intelligence is the absence of human errors that can arise due to various factors which might decrease the accuracy of results. Rebound hammer (RBH), Windsor probe penetration (WPP) and ultra-sonic pulse velocity(USV) are the non-destructive testing that has been employed to assess the concrete compressive strength. The use of multiple non-destructive testing methods than single testing methods has improved the accuracy of prediction model. Further the results of different combination models were compared using coefficient of determination which is a commonly used statistical parameter for prediction accuracy comparison. The prediction model is in accordance with the experimental results obtained. And the accuracy of prediction indicates that support vector regression can be successfully used for predicting the compressive strength of concrete.