The Punching Shear Capacity Estimation of FRP- Strengthened RC Slabs Using Artificial Neural Network and Group Method of Data Handling

Abstract

Recently soft computing methods have been employed in most fields, especially in civil engineering, due to its high accuracy to predict the results and process information. Soft computing is the result of new scientific endeavors that make modeling, analysis, and, ultimately, the control of complex systems possible with greater ease and success. The essential methods of soft computing are fuzzy logic, artificial neural networks, and genetic algorithm. In this paper, using 74 valid experimental data, estimation of punching shear capacity of FRP-strengthened RC slabs using two powerful methods (artificial neural network and Group method of data handling) has been investigated. The maximum and minimum dimension of column cross-section, the effective height of slab, the compressive strength of concrete, modulus of elasticity of FRP bar, and the percentage of FRP bars were selected as input variables, and the punching shear capacity of the slab was selected as the output variable. Also, in order to investigate the effect of the variables mentioned above on the results, sensitivity analysis is conducted in both methods. Absolute Fraction of Variance for the two methods showed that the GMDH method had higher precision (1.73%) than the ANN method in the prediction of results.