Intelligent Modelling of Punching Shear Strength in Fibre Reinforced Polymer Concrete Slabs Using Gene Expression Programming and Machine Learning

Running Title ML prediction of punching shear in FRP slabs Abstract The increasing demand for resilient structures in earthquake-prone and high-impact areas has necessitated the accurate prediction of the punching shear strength of concrete slabs reinforced with FRP (Fiber Reinforced Polymer) and without shear reinforcement. The classical approaches, including physical testing and mathematically derived models, tend to have limitations in terms of both accuracy and speed. The present work carries out the prediction of the punching shear strength of FRP-reinforced concrete slabs by employing various machine learning (ML) techniques like decision tree, random forest, stochastic gradient boosting, and so on. The research conducted involved a thorough study through a dataset comprising experiments on slabs with different design parameters, including thickness, strength of the concrete, type of FRP reinforcement, and ratio of reinforcement, which was then used to train and evaluate the ML-based models. Out of all the models that were put to the test, the GEP (Gene Expression Programming) method attained the utmost R² value equal to 0.86, which is an indication of getting shear strength prediction more accurately than traditional models. Furthermore, the ML models were compared through statistical errors, including RMSE, MAE, and R², with the attained values being within the acceptable range. The analysis of feature importance showed that the parameters slab thickness, concrete strength, and reinforcement ratios were the key players in the determination of the punching shear strength. The present study clearly points to the machine learning potential in reducing the effort and time in punching shear strength prediction, the outcome being better design codes and standards for FRP-reinforced concrete slabs in the future. Keywords Punching shear strength; FRP-reinforced concrete; machine learning; gene expression programming; slab thickness; reinforcement ratio Article Type Research Article Subject Area / Discipline Structural Engineering / Civil Engineering Funding Information None Conflicts of Interest The authors declare no conflicts of interest. Ethics Approval Not applicable. This study does not involve human participants, animals, or sensitive personal data. Data Availability The data supporting the findings of this study are confidential and are therefore not publicly available.