Application of Six Metaheuristic Optimization Algorithms and Random Forest in the uniaxial compressive strength of rock prediction

The uniaxial compressive strength (UCS) is one of the most important parameters for judging the mechanical behavior of rock mass in rock engineering design and excavation such as tunnels, subways, drilling, slope and mines stability. However, it is difficult to obtain UCS accurately and quickly in traditional experimental operations. Therefore, prediction of the UCS of rock is of high practical significance in reducing calculation time and improving the precision of results. In this investigation, estimation and prediction of the UCS obtained from various rock in the laboratory on the base of artificial intelligence algorithms and empirical approaches were carried out. A total of 226 rock samples were selected to generate a dataset including five individual parameters, Schmidt hardness rebound number (SHR), P- wave velocity ( V p ), point load strength (Is (50) ), porosity (n), and density (D). The artificial neural network (ANN), kernel based extreme learning machine (KELM), support vector regression (SVR), empirical equations and a hybrid model Slime Mould Algorithm-based random forest (SMA- RF) were developed to predict the UCS. Four performance indicators named the root mean square error (RMSE), the determination coefficient (R 2 ), the mean absolute error (MAE) and the variance accounted for (VAF) were utilized to evaluate the performance of all models in forecasting the UCS of rock. The results of performance comparison demonstrated that the SMA- RF model has the highest values of R 2 (train: 0.9907 and test: 0.9705) and VAF (train: 99.0713 % and test: 97.0753 %), the lowest values of RMSE (train: 4.1478 and test: 7.7824) and MAE (train: 3.0096 and test: 5.8532) among the other models. The research in this study provides an effective attempt to further improve the accuracy of UCS prediction. • Application of six emerging Metaheuristic Optimization Algorithms and RF model in predicting the uniaxial compressive strength (UCS) of rock. • A comprehensive dataset of 226 rock samples with five properties was generated on the base of the four published articles. • The TSO-RF represents the best performance in UCS prediction among all hybrid RF models and other AI models.