Abstract: In order to overcome the difficulty of obtaining ground stress in real time during the construction of tunnels using the drilling and blasting method, this paper developed an impact rotary type high-precision synchronous acquisition equipment for drilling parameters and a three-dimensional controllable stress loading platform. 12 sets of drilling tests with different surrounding rock ground stress states were carried out to systematically reveal the response laws between drilling parameters and ground stress, and between ground stress and wave velocity. On this basis, a five-dimensional drilling feature-wave velocity sample library of 6000 multi-conditions was constructed, and a bagging decision tree integration model was designed to realize the intelligent prediction of three-dimensional wave velocity. The principal stress magnitude was inverted based on the acoustic elastic theory. At the same time, a principal stress direction analysis algorithm based on wave velocity ellipsoid regression was proposed. The results show that the prediction accuracy of the three-dimensional wave velocity model is 84.44%-88.21%, and the principal stress magnitude inverted using the predicted wave velocity is in good agreement with the loading value. The maximum principal stress direction error of the wave velocity ellipsoid method is ≤0.1°, and the errors of the other principal stress directions are ≤18.2°. The research results provide new ideas for the rapid and intelligent analysis of the ground stress field of the tunnel surrounding rock, and can provide important technical support for the construction safety and support optimization design of high ground stress tunnels.