Adjustments of error by neural networks for the shear stress carried by the stirrups of a beam-column joint

摘要

The application of neural networks in the form of parameter predictions to assess the shear stress carried by the stirrups of reinforced concrete beam-column joint under axial load and biaxial bending has been considered. Computation algorithms in the form of numerical analysis were performed on the beam-column joint to simulate existing experimental data. The focus of this paper is to reconstruct existing experimental data by evaluating several parameters and establishing valid mathematical relationships based on neural networks that are in close agreement with existing relationships based on experimental results. Adjustments of error due to lateral stirrup spacing were carried out using neural network techniques to enhance the values of shear stress carried by the stirrups within the joint. The procedure demonstrates the capability of the back propagation algorithm with supervised learning to enhance the lateral spacing by reducing the percentage errors present in analysing experimental results.