The Research Of Security Evaluation In Highway Tunnel Structer Base On The Neural Network

There are certain risks in the construction of highway tunnel for its hidden nature, uncertainty in geological conditions and other environmental influences. In order to ensure that the tunnel construction can realize successfully, it is necessary to research on the assessment and management of these risks. By analysis and synthesis judgments of the measured data, we can perfect our design and take effective measures in face of these risks.This paper is based on the Xishan tunnel engineering of Taiyuan-Gujiao highway, taking the surrounding rock pressure of highway tunnel and the stress deformation of supporting as the research objects, selectly using fuzzy comprehensive evaluation and neural network evaluation method to study on the safety assessment of the tunnel construction structure.First of all, we made a brief review on the current security situation and the problem of tunnel construction in our country, it is concluded that the necessity of safety assessment. Then we analyzed the basic theory of the safety evaluation as well as the method and procedures used in safety evaluation. Based on monitoring various sensors and data collectors which are installed in special points of the tunnel, we can obtain the pressure of surrounding rocks data, the stress of deep foundations data and the subsidence of vaults etc, then the data is transmitted to the central computer through wireless network for analyzing and comprehensive judgment. Therefore, the monitoring measurement quality become an important index of measuring whether the design and construction is reasonable or not.In this paper, we established the safety evaluation model on theory of fuzzy comprehensive evaluation and BP neural network to analyze the Xishan tunnels of the Tai-Gu highway. By analyzing the evaluation results of the two methods, we found that the fuzzy comprehensive evaluation can't perfect reflect the practical safety state of tunnel due to many uncertain factors, but the effect is better which using neural network evaluation method to analyze the tunnel safety assessment. Finally, in order to verify the feasibility of the evaluation method, we analyzed the monitoring data according to the established model, there has a very important significance to grasp the dynamic changes rule of surrounding rocks and the safety status of deep foundation. Thereby we can take corresponding measures timely in face of risks and provide effective reference for the safe construction and operation in the tunnel, which has a certain engineering value in practical use.