Evaluation of artificial neural network, neuro-fuzzy and multivariate regression modelling for prediction of concrete compressive strength via point load test

Nowadays, the determination of in place concrete strength is more considered. The necessity of in situ testing can be expressed in a variety of applications such as structural change or development, quality control, strength evaluation and concrete performance. In this study the performance of Artificial neural network, Adaptive neuro-fuzzy and Multivariate regression function for measuring the concrete compressive strength with point load method are studied. Also, a computational relation is presented based on multivariate regression method for prediction of concrete compressive strength with point load method. The results indicated that the neural network, neuro-fuzzy and regression models are suitable in prediction of concrete strength with point load method. The correlation coefficients of neural network, neuro-fuzzy and nonlinear regression models were obtained 0.9412, 0.8244 and 0.8938 respectively. This indicates less error and as a result better accuracy and performance of neural network in prediction of concrete strength with point load method. The results of this study showed that there is good agreement between concrete compressive strength test with soft computing methods and real observations. The proposed method of this study, in addition to ease, reduces the time of evaluation of in situ concrete strength and reduces the cost of laboratory studies.