AI-Driven Shear Capacity Model of Steel Studs in Composite Structural Systems

In composite steel-concrete structures, shear connectors in the form of headed steel studs are commonly utilized to transfer longitudinal shear force developed at the interface between the two materials. To overcome the shortcomings of design codes, which frequently understate shear capacity and fail to take advantage of sophisticated computational methods, this paper presents an optimization attempt to estimate the shear strength of headed steel studs utilizing the Grey Wolf Optimizer (GWO) technique using MATLAB software. Data from 234 experimental tests are employed to identify and highlight key input parameters influencing the shear strength of headed steel studs. These key parameters include concrete compressive strength (f’c), diameter (D), and tensile strength of the steel stud shank (fu). After identifying and examining the limits of the experimental data, the proposed model has been developed using about 80% of the mixed raw dataset. The remaining 20% of the raw data is utilized to validate the proposed model. The predicted shear strength of headed steel studs closely matched the experimental results. This research offers an innovative strategy to measure the steel stud's shear capacity employing GWO, showing the current code's limitations. The GWO model showed excellent accuracy in predicting the shear strength with an R-value of 0.9922, indicating that the predicted value is in good agreement with experimental observations. Interestingly, the model's mean absolute error with 100 wolves in the GWO method was only 7.51%, showing the proposed model provides an improvement in shear capacity forecasting for practical structural engineering applications.