Predication of the residual axial load capacity of CFRP-strengthened RC column subjected to blast loading using artificial neural network

Abstract

In this study, two genetic algorithm optimized backpropagation neural networks (GA-BPNN) were established to predict the ratio of residual axial load capacity to the maximum axial load capacity (referred to as RCI hereafter) of the non- and CFRP-strengthened RC columns based on a huge amount of simulation data. The first one can be used to predict the residual axial load capacity of the damaged non- and CFRP-strengthened RC columns induced by blast load with the input of several parameters including column dimensions, concrete strength, transverse reinforcement ratio, longitudinal reinforcement ratio, axial load ratio, CFRP stiffness, carbon fiber strength, peak pressure and impulse of the blast load. Therefore it can be used for the blast-resistant design of non- and CFRP-strengthened RC columns. The input variables of the second GA-BPNN were changed to be the ratio of residual mid-height deflection to the column height after the explosion, column dimensions, concrete strength, transverse reinforcement ratio, longitudinal reinforcement ratio, CFRP stiffness and carbon fiber strength. Since the input variables of the second GA-BPNN could be easily derived after the explosion, thus it could be used for the rapid damage assessment of RC columns. Damage assessments for three non- and CFRP-strengthened columns were also conducted using the first GA-BPNN.