Machine learning approach to analyze the effect of the micro silica on mechanical properties of the concrete at elevated temperature

Concrete, a construction material worldwide, undergoes continuous refinement to improve its mechanical attributes. Traditionally, evaluating the influence of micro silica on concrete properties necessitates extensive experimental scrutiny, which is often resource-intensive, and susceptible to external variables. In this study, concrete formulations incorporating cement and varying amounts of micro silica (MS) were developed. Mechanical properties were assessed across different micro silica compositions after 28 days. The study observed that as the MS content increased from 10 to 50 Kg/m³, there was an initial increase in strength followed by a subsequent decrease. Additionally, the impact of elevated temperatures (ranging from 27 to 700 °C) on concrete strength was investigated. Results indicated a strength increase up to 100 °C, followed by a decline beyond 300 °C. Furthermore, artificial neural network (ANN) analysis was employed to evaluate the obtained data, revealing a strong correlation (R² = 0.95) between the model predictions and the compressive strength mechanical properties. In the training phase, counter propagation neural network (CPNN) achieved remarkable success, with all mechanical properties of compressive strength, flexural strength and split tensile strength surpassing a threshold of 0.95. Although CPNN showed exceptional performance in training, ANN displayed superior generalization abilities, as evidenced by its outstanding performance on the testing dataset.