Study On The Confidence Of Blasting Control Technology Of Tunnel Passing Underneath Buildings

Recently,with the sustained growth of the national economy,the process of urbanization is deepening,as a result,the construction of urban transportation infrastructure is also developing rapidly,and more and more tunnel projects are being built beneath buildings or structures in urban areas.Limited by engineering geology,economic factor and other factors,drilling and blasting method is still used for the excavation of tunnels in the city.However,vibration induced from blasting has a serious influence on surroundings,it causes vibration for buildings above the tunnel,and affects normal life of people nearby.Therefore,it is very important to study the blasting vibration control technology of the tunnel beneath buildings.This paper is based on project of Dalian South Binhai Road Tunnel crossing beneath Tongxiangshanzhuang villa,regression analysis and confidence research of field monitoring data of blasting vibration are carried out,the main research results are as follows:(1)For field monitoring data of blasting vibration used in this paper,fitting line obtained from regression analysis merely reflects the average degree of data points distribution,nearly 50%of data points are distributed over the fitting line,and the rest are distributed under the fitting line.In the actual project,if the fitting line is used as the control criterion of blasting vibration,blasting vibration velocity must have a great possibility(nearly 50%)exceeding the control criterion,thus causing many problems.Therefore,it’s necessary to introduce the 95%confidence line,which can help to control the 95%of blasting under the control criterion.As a result,it not only ensures safety,but also is not too conservative to influence construction,so that the blasting excavation of the tunnel beneath buildings can be carried out smoothly.(2)Compared with conventional analysis and prediction models,the model based on Bayes Theory although is complex in both principle and model implementation,it shows a high precision in predicting the confidence line of the blasting vibration,especially in the case that the sample correlation is not high.(3)The sample data has an influence on the result of regression analysis and confidence research.On the one hand,the number of sample data affects the precision of the results,generally speaking,the more the number of samples,the more precious the predicting result will be.On the other hand,the characteristics of the sample data itself,such as correlation,degree of dispersion,also have an impact on the regression analysis and confidence prediction accuracy.(4)In the previous study,BP neural network method showed high accuracy in nonlinear regression analysis,but because of its prediction ability is influenced by the sample characteristics used for training,the BP neural network method gives a poor prediction for the sample data used in this paper.However,although the vibration velocity predicted by BP neural network is not precise,it still has an advantage over fitting line method when comparing the prediction accuracy,and it still meets the confidence limit control requirements.