Gaussian Process Model of an Advanced Surrounding Rock Classification Based on Tunnel Seismic Predictions

In order to improve the prediction methods of geological hazards in the front of a tunnel face. Gaussian process machine-learning is proposed and the classification model is established for advanced surrounding rock classification based on TSP-203 system. This approach extracts the useful information from the detection results of TSP-203, and establishes the quantitative index system of surrounding rock classification. Then the unknown nonlinear mapping between the detected geological information and the actual surrounding rock level is established by Gaussian process classification model (GPC) ultimately. To verify the feasibility and superiority, GPC model is applied to Jinpingyan Tunnel on Chenglan Railway in China and the results show that GPC performs a higher prediction accuracy than support vector machine (SVM) and grey clustering method in predicting surrounding rock level, providing a new approach for advanced surrounding rock classification.