Intelligent prediction of tunnel surrounding rock advance classification in high altitude and high seismic intensity area and its engineering application

Abstract

In order to improve and optimize the advance classification and prediction method of tunnel surrounding rock, a prediction method based on Tunnel Seismic Prediction (TSP) and Probabilistic Neural Network (PNN) is proposed. Based on the characteristics of science, maneuverability and representativeness, several factors that greatly affect rock mass classification are selected as evaluation indices based on analysis of numerous TSP data, establishing an advance classification index system for surrounding rock, and designing the “Advance classification and prediction system for surrounding rock” to predict the classification. Engineering application of Jinpingyan Tunnel of Chenglan Railway in high altitude and high intensity area of China is taken as a case study, and proved that the evaluation indices are easy to obtain and the evaluation results are accurate and reliable, and compared with Back Propagation (BP) neural network prediction results, the results show that PNN has some advantages in predicting the calculation speed of surrounding rock classification, the ability to add samples and the classification accuracy in practical engineering applications. The PNN-TSP method can be further used for other tunnel engineering.