Data-centric approach for predicting structural behavior of ultra-high-performance concrete beams

The long-term improved durability and mechanical properties of ultra-high-performance concrete (UHPC) are widely recognised. As a result, it has applications in structural construction, with the highest efficiency for beams. Predicting the structural behaviour of UHPC beams is difficult due to numerous interactions between material properties, geometric configurations, and loading conditions with the environment. Traditional modelling methodologies fail to address this complexity, resulting in erroneous and inefficient designs. This study proposes a data-driven technique for predicting the structural performance of UHPC beams using the Intelligent Cuckoo Search Algorithm-Refined Random Forest (ICSA-RRF) model. An adequate dataset covering various aspects of UHPC beams, including stress-strain characteristics, load-bearing capacities, and other failure mechanisms, was created. Some preprocessing techniques, such as Z-score normalization and Principal Component Analysis (PCA), were employed to extract features and reduce dimensionality to achieve the highest data quality and prediction model performance. The ICSA-RRF model was utilized to identify essential factors and then develop a robust predictive framework. The model's performance was evaluated using main performance indicators, including a coefficient of determination R-squared (R²) of 0.98, RMSE of 2.19, and MAE of 1.98. Such measures demonstrate the precision and accuracy with which UHPC beam performance is analysed. Thus, the proposed ICSA-RRF model reduces the design process by avoiding lengthy physical testing, resulting in more optimized safe designs. Furthermore, reliable prediction of a structure's behaviour under various conditions can assist engineers and construction professionals in generating more informed decisions. Under these circumstances, the ICSA-RRF model demonstrates an intelligent technique for correct forecasting of UHPC-behaviour beams, hence increasing efficiency and dependability in design structuring. Future studies can expand this approach to other structural elements and incorporate additional variables, such as climatic conditions and long-term performance data, to improve predictions even more.