Parametric FEM Analysis of Cellular Composite Steel–Concrete Beams and Deflection Prediction Using Least Squares Support Vector Machine Optimized with Metaheuristic Algorithms

Cellular composite steel–concrete beams (CCSCBs) are often utilized in tall structures. These openings are used to accommodate facilities like ducts and ventilation pipes. However, parameters related to the design of a cellular composite steel–concrete beam (CCSCB), such as diameter, spacing, etc., lack in-depth analysis. Therefore, in this study, finite element method (FEM) analysis was executed on a set of 66 models with opening diameters of 0.25 D (50 mm), 0.5 D (100 mm), and 0.75 D (200 mm). The spacing-to-diameter ratio of the opening (S/Do), which ranges between 1.08 and 3, was adopted. The primary objective of the present investigation was to evaluate the load-bearing capacity, considering the opening area and the ratio between the opening diameter and the available space in the web. Furthermore, using the FEM analysis results, a hybrid intelligence model was proposed to predict the deflection of these beams accurately. Accurate assessments of deflection behavior can significantly contribute to preventing unanticipated and severe structural damage and its accompanying adverse consequences. The load-deflection behavior was effectively estimated using a least squares support vector machine (LSSVM) model, enhanced by optimization through two metaheuristic techniques: particle swarm optimization (PSO) and Harris hawk optimization (HHO). The results obtained from this evaluation provide evidence that the suggested LSSVM-HHO model exhibits a high level of accuracy in predicting deflection behavior. This model achieved a Nash–Sutcliffe efficiency (NSE) of 0.929 and a coefficient of determination (R2) of 0.93. The proposed LSSVM-HHO model shows excellent potential as a practical approach for predicting the deflection behavior of cellular steel–concrete composite beams.