Innovative models for prediction of compressive strength of FRP-confined circular reinforced concrete columns using soft computing methods

There are several methods for predicting experimental results such as empirical methods, elasticity and plasticity theory. Among these methods, the use of soft computing has been expanded due to good capabilities and high accuracy in predicting the target. Soft computing contains computational techniques and algorithms to provide useful solutions to deal with complex computational problems. In this study, three methods including Artificial neural networks, Group method of data handling and Gene expression programming are utilized to predict the compressive strength of columns confined with FRP. Total of 95 experimental data were selected to form the model. The height of the column, the compressive strength of unconfined concrete, the elastic modulus of FRP, the area of longitudinal steel, the yield strength of longitudinal steel and confinement pressure provided by FRP and transverse steel were considered as input parameters, while the compressive strength of FRP-confined columns was considered as the target. The proposed methods are compared with the existing models and provide great accuracy in predicting the results. Among the utilized methods, the ANN model showed the highest accuracy.