Abstract: To address the technical challenge of rapidly obtaining in-situ stress during tunnel construction, an in-situ stress estimation method for tunnels based on measurement while drilling (MWD) data and acoustic information is developed in this study using particle swarm optimization-support vector regression (PSO-SVR). This method integrates MWD data and acoustic wave information to estimate in-situ stress. By utilizing multidimensional MWD data from a drilling jumbo (including impact pressure, feed pressure, rotation pressure, drilling speed, etc.), together with known compressive strength, elastic modulus, and Poisson′s ratio, a real-time MWD prediction model based on PSO-SVR is established, enabling efficient acquisition of rock mechanical parameters. The cross-hole wave velocity vector information is collated, the elliptical characteristic parameters are obtained through wave velocity ellipse fitting, and the magnitude and direction of in-situ stress are calculated based on the predicted mechanical parameters of rocks by using the acoustic-elastic theory, thus realizing the rapid estimation of in-situ stress. To verify the accuracy and applicability of this method, a field testing program is designed and applied to a plateau railway tunnel. The test results show that the average errors of magnitude and orientation of in-situ stress compared with conventional testing methods are 7.27% and 12.51%, respectively, and the time required for a single test is <2 h, substantially improving the testing efficiency.

Key words: tunnel engineering, in-situ stress, particle swarm optimization-support vector regression, acoustic-elastic theory, measurement while drilling data, ultrasonic velocity ellipse