Coupling Beams’ Shear Capacity Prediction by Hybrid Support Vector Regression and Particle Swarm Optimization

In structures with reinforced concrete walls, coupling beams join individual walls to produce a rigid assembly that withstands sideways forces. A precise forecasting of the critical shear capacity is essential to avoid early shear failure and attain the desired ductility performance of coupled shear wall systems in earthquake design. This paper examines the ability of Support Vector Regression (SVR) in predicting the shear performance of coupling beams. SVR is a distinguished machine learning regression method that has been positively utilized in former works to forecast the performance of several structural members. Nevertheless, the capability of this regression method deeply relies on picking its best hyperparameters. To handle this, a heuristic optimization procedure named Particle Swarm Optimization (PSO) was merged with SVR to select the optimal hyperparameters. The data of RC coupling beams collected from the previous works were utilized to build the proposed model. Several performance metrics, including RMSE, R2, and MAE, were employed to compare the performance of the optimized model against a baseline SVR model and previous approaches. Analytical results indicate that the new optimized prediction model can assist civil engineers in designing RC coupling beam structures more effectively and outperforms existing models in predicting the shear strength of such beams.

This creative work has been published under the Creative Commons Attribution 4.0 International (CC-BY 4.0) license which allows copying, and redistribution as well as adaptation of the original work provided appropriate credit is given to the original author and the conditions of the license are met.