Anticipation ofMid-Span Deflection atUltimate Loadfor Concrete BeamsStrengthenedby FRP BarsUsing Artificial Neural Networks

Abstract

A mid-span deflection of concrete beam strengthened by FRP bars has beenanticipated. A case of simply supported beam loaded by two point loads has been simulated by using artificial neural networks (ANNs) which are involved in MATLAB package, version 9.0.0.341360 (R2016a). The proposed model presupposes60 beam specimens to collect the required data for the neuralpattern. A set of 8 input variables was selected to construct the proposed neural pattern; these are, beam dimensions, concrete specifications, and FRP properties. While the mid-span deflection at the ultimate load, will be the output inconstant. The use of 9 nodes in the hidden layer is active in predicting the mid-span deflection. A comparison between the gained results and the past experimental data shows that the proposed neural patterngives a reasonable anticipation with an overall error of 8.2 %.