Precise evaluation of rock mass quality is crucial for the design and building of underground structures like tunnels and subterranean areas. Despite being one of the most commonly utilized methods in rock engineering, the conventional estimation of the Q classification system demands significant field studies and lengthy, expensive lab tests. This study aims to present a quick, cost-effective, and non-invasive technique for indirectly estimating the Q-value by modeling the correlation between compressional (Vp) and shear (Vs) wave velocities alongside the Q classification index through the use of metaheuristic algorithms. Thus, three predictive models were developed employing simulated annealing (SA), particle swarm optimization (PSO), and a hybrid algorithm that combines PSO and SA. The algorithms underwent training and testing with actual data sourced from a particular case study tunnel in southwestern Iran, which included various types of rock. The effectiveness of the models was evaluated using various statistical measures, including R², RMSE, MAE, MAPE, A20, IOA, and IOS. Additionally, score evaluation and REC curves were utilized to offer a more thorough comparison. The modeling procedure was conducted in two distinct scenarios: utilizing Vp and Vs individually as the exclusive input for the models. In each instance, the hybrid PSO-SA model exceeded the performance of the other two models. Comparative findings revealed that the PSO-SA model using Vs as input reached an R² of 0.9629 and an RMSE of 0.6342, demonstrating greater accuracy compared to Vp (R²=0.9504, RMSE=0.7232). All metrics favored the Vs-based model, with the exception of the MAPE index, which performed slightly better with Vp. These results indicate that Vs might act as a stronger predictor of the Q-value. In general, employing computational intelligence models, particularly hybrid metaheuristic algorithms such as PSO-SA, provides a dependable, effective, and precise approach for assessing the quality of the rock mass.
