New Assessment Method on Shear Resistance of Perfobond Shear Connectors in Steel-Concrete Composite Structure

The behaviour of shear connectors will have a major impact on the global behavior of steel-concrete composite structure. Perfobond shear connector has been developed during last two decades and has become popular due to their advantageous properties. Longitudinal shear strength is considered as a major constraint in the design of composite structure and it can be assessed by expensive and time consuming experimental techniques. Longitudinal shear resistance of perfobond shear connectors is depends on some design parameters. In this paper, considering its powerful prediction ability on nonlinear problem, artificial neural networks was induced to investigating and a new intelligent evaluation method on shear resistance of perfobond shear connectors was proposed. Choosing the diameter of holes, the yield strength of steel plate, concrete compressive strength, the ratio of transverse rebar, depth of steel plate and thick of steel plate as input values, Back Propagation Neural Networks (BPNN) model was developed for determination of the longitudinal shear resistance of perfobond shear connectors. It is demonstrated that, with the same design parameters as test specimens, the longitudinal shear resistance generated by the BPNN model is quite close to test result after proper training of the BPNN. Furthermore, Instead of three-dimensional FEM Analysis or Push-out test, the BPNN model is computationally efficient tool used to predict shear resistance of perfobond shear connectors in different parameters..