Punching Shear Strength Estimation of FRP‐Strengthened RCC Slabs Using Hybrid Deep Learning and Meta‐Heuristic Optimization Approach

ABSTRACT Reinforced concrete flat slabs (RCFS) can be architecturally flexible because of the regular thickness and support on columns without the use of conventional beams. Nevertheless, they have been found to be prone to punching shear failure, which has posed some safety issues with respect to the design of the structures, especially under extreme loading. This paper introduces a new predictive model to calculate the punching shear capacity of strengthened RC flat slabs with fiber reinforced polymers (FRP), which is done through data‐savvy approaches. This methodology involves a systematic preprocessing of the data, statistical feature extraction, feature selection using a Reptile Ring Toss Hybrid Optimization Algorithm (RRTHOA), and finally the prediction by means of a fused deep learning architecture that integrates ANN, capsule networks (CapsNet), and recurrent neural networks (RNN). The proposed model performed better as it had an MAPE of 0.8208, MAE of 0.8317, MSE of 4.6072, RMSE of 1.7734, correlation coefficient of 0.9926, and overall accuracy of 98.87%. The findings of this study suggest that the hybrid optimization deep learning model could achieve better estimation results in the punching shear capacity than the conventional methods. The results demonstrate the value of the proposed model to support the design, analysis, and repair of RC slabs strengthened with FRPs, thereby making the structural engineering practice even more reliable and efficient.