Research On TBM Tunneling Performance Prediction And Cutter Wear Based On Nonlinear Regression And Neural Network

TBM tunneling performance prediction has always been a hot issue in the field of TBM research.The tunneling performance mainly includes the aspects of TBM penetration,advance rate,construction speed,utilization rate,and cutter wear.Its research and forecasting is of great significance for project duration and cost estimation,construction method selection,and poor ground warning.For this,penetration rate prediction is the basis for the prediction of advance rate and construction speed.Taking the Xinjiang EH tunnel project as the background and collecting the TBM field excavation data,the orrelation analysis was performed for the excavability indexes FPI and TPI with geological parameters and operational parameters based on the traditional BQ rock grading method.Multivariate regression equations for FPI,TPI,uniaxial compressive strength(UCS)and single-cutter thrust(F_n)under different grades of surrounding rocks were set up using nonlinear regression method,then the TBM penetration prediction model was established using nonlinear regression method under different surrounding rock category.Based on field measured data,the accuracy of the FPI penetration model and the TPI penetration model under different surrounding rocks are compared and analyzed,and a penetration prediction model with better accuracy is obtained.Reliability testing of the model were performed at other Xinjiang EH tunnel section.This model is suitable for tasks such as TBM tunnel construction period and cost estimation,bidding and contract negotiation,and selection of construction schemes.Further,on the basis of the above-mentioned correlation analysis of the rock excavability index with the tunneling parameters and geological parameters,and the study of the penetration prediction model,in order to compensate for the application of the traditional BQ method for rock classification in TBM construction and the prediction of tunneling performance,and to obtain a reliable and universal prediction model of TBM penetration,the grading method of surrounding rock in accordance with the characteristics of TBM construction was presented based on two 7-meter diameter TBM field excavation data considering the TBM tunneling performance and construction risk.Then,selecting single-cutter thrust and uniaxial compressive strength of surrounding rock as basic indicators,a TBM penetration prediction model under different surrounding rock grading was established using genetic algorithm to optimize BP neural network.Xinjiang EH project and Xinjiang ABH project were used to test the reliability and universality of the model,and the accuracy of the model met the actual engineering requirements.The model has good universality and can be applied for TBM construction period estimation in project planning,real-time construction period prediction during construction,TBM tunneling state monitoring and waring of bad geology.Finally,In view of the late application of domestic cutters in hard rock TBM projects,the issues on cutter wear and consumption were studied based on site geological data and cutter wear data from the Xinjiang EH tunnel project.The wear rule of cutters,and the total number of consumed cutters per meter advance,the normal wear consumption number of cutters per meter advance,cutter wear height per meter advance,and the average available advance length of reach cutter at different installed position under the three types of lithology are given.Aiming at the problem of cutter selection under different geological conditions,on-site test tests and analyses were conducted to compare the performances of two domestic manufacturers'cutters under different surrounding rocks,which provides important references for the future cutter cost estimation,reasonable stocking of cutters,and cutter selection.