Research On The Prediction Of Soft Surrounding Rock Tunnel Deformation Based On Neural Network Algorithm

The strategy of "one belt,one road" strategy is new strategies proposed by our country in recent years.So the implementation of the strategy will effectively bring about economic development in the western region of China.The infrastructure construction in the western region is particularly important,especially the construction of the transportation network system.However,the central and western regions of China are mountainous areas with complex geological conditions.At the same time,tunnel engineering has become an important part of the Western transportation system.Therefore,the bridge and tunnel problem has become a key problem restricting the development of the Western transportation system.In the process of highway and railway tunnel construction in the western region,it is inevitable to encounter high ground stress and deep buried soft surrounding rock.Because of its complex geological situation,the surrounding rock often loses stability,threatening people’s life and property safety.In this paper,based on the analysis of the deformation of a soft surrounding rock highway tunnel in northwest China,BP,RBF and GRNN neural network algorithms are used to predict the vault deformation of soft rock tunnel.This paper compares the prediction effect of neural network for the tunnel and evaluates the prediction accuracy of neural network.Finally,the neural network algorithm with high prediction accuracy is selected for promotion and application.The monitoring data of vault,arch waist,arch foot and step deformation on site were analyzed and sorted out.Seven parameters were selected as the characteristic parameters and vault deformation as the target parameters.Then,relevant neural network algorithm was designed and compiled to predict the deformation of the tunnel’s weak surrounding rock vault.The whole research work achieved the following research results:(1)This paper analyzes the engineering geology,hydrogeology,burial depth,lithology and other factors of the soft surrounding rock highway tunnel,and obtains the internal relationship between the deformation of the soft surrounding rock and many factors based on the deformation mechanism of the soft surrounding rock.(2)Based on the analysis of the deformation of various parts of the tunnel(cave)in the soft surrounding rock,the relationship between various deformation factors was determined and summarized.Seven deformation parameters with strong correlation were selected as the characteristic parameters and one deformation parameter as the target parameters to predict the deformation of the soft surrounding rock.(3)BP,RBF and GRNN neural network algorithms were designed and compiled to predict the deformation of the tunnel’s weak surrounding rock vault.By comparing their prediction accuracy,it was found that the prediction accuracy of RBF and GRNN neural network algorithm was significantly higher than that of the standard BP neural network algorithm.(4)In view of the shortcomings in BP,RBF and GRNN neural network algorithms,genetic algorithm and particle swarm optimization algorithm are used to optimize the neural network algorithm.The prediction accuracy of the optimized neural network algorithm has been significantly improved,especially the BP neural network algorithm,which has the most obvious improvement effect.