Accurate evaluation of rock mass quality plays a vital role in the design, stability analysis, and construction of underground excavations such as tunnels and caverns. The Q classification system is one of the most widely used empirical approaches in rock engineering. However, determining the Q value traditionally demands extensive field investigations and laboratory testing, which are often time-consuming, expensive, and subjective. This research aims to develop a fast, reliable, and non-destructive method to indirectly estimate the Q index by establishing its correlation with the compressional (Vp) and shear (Vs) wave velocities of rocks. To achieve this, three computational intelligence models including simulated annealing (SA), particle swarm optimization (PSO), and a hybrid PSO–SA algorithm were developed and trained using actual data obtained from the Beheshtabad water transfer tunnel in southwestern Iran, a site characterized by diverse geological formations. The predictive capability of each model was assessed using several statistical criteria, including R², RMSE, MAE, MAPE, A20, IOA, and IOS, alongside supplementary score and REC curve analyses. The modeling was performed in two scenarios: one using Vp and the other using Vs as the sole input variable. In both cases, the hybrid PSO–SA algorithm exhibited superior performance. When Vs was used as input, the model achieved an R² of 0.9629 and an RMSE of 0.6342, outperforming the Vp-based model (R²=0.9504, RMSE=0.7232). The results suggest that Vs serves as a more reliable indicator for estimating rock mass quality under the studied conditions. Overall, the proposed hybrid metaheuristic model offers a robust, cost-effective, and non-invasive framework for predicting the Q value, significantly reducing the dependence on traditional field-based measurements.
