Spilt Tensile Strength of Fiber-Reinforced Recycled Aggregate Concrete Simulation Employing Tunned Random Forests Trees

A significant quantity of waste concrete is produced each year due to the demand for concrete manufacturing, which drives the yearly need for raw materials. Recycled aggregate concrete has become a viable remedy as a result. It is vulnerable to breaking and has less strength since the hardened mortar is affixed to natural aggregates, which presents a problem. The goal of this research is to employ random forests (RF) frameworks to project the split tensile strength (STS) of fiber-reinforced recycled aggregate concrete (RAC). The RF framework uses the Chimp optimization algorithm (CHOA) and artificial hummingbird optimization (ARHA) to tweak hyperparameters and select the best-performing combination. A data set including 257 data points and 10 input variables was taken from peer-reviewed published research and arbitrarily split into three phases: testing, validating, and training. The RF-AR approach exhibited high reliability, with R2 of 0.9942, 0.9824, and 0.9913 throughout the learning, validating, and assessment stages. RF-AR had higher results than RF-CH, with R2 of 0.9796, 0.9566, and 0.9694, respectively. Considering the values of the Theil inequality coefficient (TIC), RF-AR depicted the lowest values at 0.0128, 0.0213, and 0.0171 concerning 0.0241, 0.0333, and 0.0318 related to RF-CH for the train, validation as well as test phases, in that order. The RF-AR strategy performed better, even if the RF-CH method was dependable in forecasting the STS of fiber-reinforced RAC, according to the previously stated reasoning and the data.