In-situ Data Analysis Of Tunneling Engineering And Calculation Of Total Loads Of The Equipment

In the data era,more and more engineering equipments establish the calculation model of key working parameters through the analysis and mining of the condition monitoring data to achieve optimal control.How to extract the rich influence relationship and rule of the parameters contained in the engineering data by effective data analysis method is a research hotspot in this field.Full section tunnel boring machine is a kind of large engineering equipment specially used in tunnel construction.Because the underground construction environment is changeable and the equipment has the characteristics of heavy load,it’s control in the process of tunnel construction has a very high requirement.The total tunneling load(including the total thrust and the total torque)is one of the core parameters in the tunneling process.When the equipment is excavated under various geological conditions,the effective analysis and calculation of load parameters is an important foundation for safe and efficient construction.With the improvement of tunneling equipment manufacturing technology,it has been realized to obtain a wealth of in-situ data from the process measurement and environmental monitoring,which lays a foundation for the tunneling loads calculation based on in-situ data analysis.In this paper,the tunneling total load is taken as the research target,and based on the parameter modeling method combining mechanical dimensional analysis and data mining,a basic quantity selection scheme of the dimensional analysis of tunneling total load is designed,and the mechanical failure characteristics of different tunneling geology are analyzed.By means of grouping and screening,the basic parameters are determined from three aspects: data characteristics,dimensional characteristics and mechanical properties.The dimensionless parameter sets of the total tunneling loads under three geological conditions of soil,hard rock and soft rock are constructed respectively.On the above basis,based on Lasso regression,data training and learning of six engineering projects of three types of geology are carried out,and the quantitative relations among the main dimensionless quantities affecting the total tunneling loads are identified to calculate the total thrust and torque.In the calculation process,aiming at engineering data preprocessing,a method of outlier identification combining difference method and box diagram is designed,which can identify abnormal data and avoid deleting the maximum or minimum values which deviate from the overall data mean value but show regular changes.The results show that the predicted values are basically consistent with the measured values on the independent test set.Based on the calculated results,the influence relationship between the relevant parameters of the total tunneling loads under different geological conditions is analyzed.In order to further analyze the nonlinear relationship between the dimensionless influence parameters,the total tunneling loads of the project is calculated based on the symbolic regression algorithm.Through the model training method combining random forest feature selection and symbolic regression algorithm,the modeling and prediction of total tunneling loads under three kinds of geological conditions are completed,and the efficiency of solving symbolic regression is improved.Through the analysis of the calculated results and the comparison with the Lasso regression results,the influence of the quantities on the total tunneling loads under different geological conditions is discussed.The calculation results of the two algorithms show that the dimensionless quantities obtained in this paper can effectively reflect the total tunneling loads under different geological conditions,which can provide a reference for the load prediction and control of tunneling equipment.