In-situ stress field is an important factor to be considered in the investigation and design of long, deep and buried tunnels. In order to reasonably determine the initial in-situ stress field of long, deep and buried tunnels, based on the measured in-situ stress data and deterministic numerical simulation method, through the self-learning function of BP neural network and the global optimization ability of genetic algorithm, the in-situ stress state of Yinheshan tunnel is predicted and simulated nonlinearly, and the results of nonlinear inversion are compared with those of multiple regression analysis. The results show that the geostress field is highly nonlinear. Compared with the multivariate linear regression analysis method, the average deviation between the results obtained by the nonlinear inversion method and the measured values is 20.01% (SH) and 19.8% (Sh), respectively, which is obviously smaller than that obtained by the multivariate linear regression analysis method.
正在翻译中..