Study On Tunneling Parameters Of Shield Machine Based On Big Data

In order to alleviate the traffic pressure brought by urbanization,subway construction has been called one of the most important application scenarios of shield machine.In actual shield tunneling projects,the control of shield tunneling parameters largely depends on the operator's experience.How to use the historical data of shield tunneling accumulated over the years before,confirm the relationship between various parameters and quantify the control of shield tunneling parameters has become an urgent problem to be solved.Taking Nanjing Yangtze River Tunnel as an example,combined with big data analysis process,shield tunneling parameters are studied through pretreatment,trend statistics,discrete statistics,regression analysis and machine learning modeling to determine the relationship between parameters and the optimal prediction model of each stratum.The research process and conclusions are as follows:(1)The auxiliary analysis system of shield tunneling parameters designed in this paper can help data analysts to complete the analysis task of shield tunneling parameters more quickly.(2)According to the statistical analysis of mean value,standard deviation,coefficient of variation and box chart,the changing trend of cutter head speed,penetration and driving speed is basically the same,and the three parameters are very sensitive to the strata containing gravel and gravel.The change trend of total thrust and mud pressure is basically the same in all strata.In the beginning and arrival stages of the project,the discreteness of each parameter is obviously larger.(3)According to the analysis of single regression and multiple linear regression of shield tunneling parameters,the single regression formula has poor prediction ability when the goodness of fit is better,which is mainly caused by over-fitting.Multivariate regression formulas have better prediction ability when the goodness of fit is good.This paper considers that the application of multivariate regression analysis in shield tunneling parameters analysis is better.From the results of multiple linear regression analysis of shield tunneling parameters,it can be seen that the tunneling speed is mainly affected by the speed and penetration of cutter head,while the mud pressure,cutter head torque and total thrust have less influence on the tunneling speed.(4)This paper uses k-nearest neighbor algorithm,classification and regression tree,random forest and back propagation neural network in machine learning algorithm to model and analyze shield tunneling parameters,and chooses the optimal model for each stratum combined with multiple linear regression model.