Abstract:The East Tianshan region, located in the northwest of China, has complex geological conditions. Therefore, it is more difficult to quickly and accurately predict the quality of surrounding rock in front of the tunnel face. There is a surrounding rock classification that can accurately and objectively reflect the basic characteristics of rock mass, which is an important reference basis for tunncel design and construction. In order to establish a method that can objectively and accurately evaluate the engineering geological environment and predict the grade of surrounding rock in the East Tianshan area, the East Tianshan Tunnel under construction is selected as the support project of this paper, and the typical geological section of the East Tianshan extra-long tunnel that has been excavated is selected. Based on the engineering geological zoning, the technical parameters of high correlation geophysical prospecting and the geophysical migration image, the machine learning training samples are composed. The training samples of deep learning network algorithm are compiled by Python language based on TensorFlow deep learning framework, and the prediction model of surrounding rock category is established. The new excavation section data are used to continuously verify and optimize the model. Finally, the model with the highest prediction accuracy is applied to the prediction of surrounding rock category of tunnels in Tianshan area. The results show that the model trained by TST offset image + geological zoning + geophysical prospecting index data set has the best effect.