Application of soft computing techniques for the prediction of splitting tensile strength in bacterial concrete

Concrete is the most common building material used in construction activities, but concrete cracks are inevitable thus is one of its major disadvantages. The major downside of the concrete is its lower Splitting Tensile Strength (STS) attributable to the micro crack. Bacteria have recently been utilized to self-heal concrete, treat cracks, and consolidate different construction materials. However, since the testing of the mechanical properties of concrete is time-consuming, involves destructive methods, poses material wastage, and is labor-intensive, an alternative precise strength evaluation technique is required to minimize effort and time. In the current investigation, various computational techniques, such as M5P, Random Forest (RF), Support vector machine (SVM), and Linear regression (LR), were used to predict the splitting strength of concrete using experimental datasets. The Pearson VII kernel function-based SVM (SVM-PUK) strategy was determined to be the most effective and accurate technique to predict the splitting strength value compared to other used models using Correlation Coefficient (CC) values based on statistical assessments, Box plot, and Taylor diagram. Results of the sensitivity analysis, among the other input variables used in this study to predict concrete splitting strength, reveal that curing time in days (T) is the most significant variable.