Intelligent Prediction Of In-chamber Earth Pressure For EPB Shield And Optimization Of Tunneling Parameters Based On Machine Learning

With the continuous advancement of China’s urbanization process,China’s subway construction is in full swing.Subway construction is mostly located underground in the core area of the city.During tunnel excavation,the control of surface settlement is relatively strict,and a little carelessness may cause significant damage to the safety of urban roads and even citizens.The most important thing to control surface settlement is to maintain the in-chamber earth pressure within a reasonable range.Therefore,the prediction and control of in-chamber earth pressure is the core problem to be solved for safe and efficient shield tunneling.However,the current prediction and control of in-chamber earth pressure mostly rely on the experience of technicians,which is difficult to form a standard and widely used.Therefore,it is of great significance to establish a scientific prediction model and regulation mechanism of in-chamber earth pressure.Based on the monitoring data of shield tunneling parameters of subway tunnel,this paper selects screw conveyor rotation speed,cutterhead torque,advancing speed,cutterhead rotation speed,flow rate of intake mud,total thrust force and in-chamber earth pressure as the analysis object,establishes the classification of soil excavatability and the intelligent prediction model of in-chamber earth pressure by using cluster analysis and neural network,establishes the optimization mechanism of tunneling parameters,and realizes the prediction and control of inchamber earth pressure,The specific research results are as follows:(1)Field penetration index(FPI)and torque penetration index(TPI)are selected as evaluation indexes to evaluate the excavatability of the soil in front of the shield.The system cluster analysis method is used to realize the identification of soil excavatability,and the reinforcement part at the head and tail of the tunnel is successfully identified.The realization of soil excavation classification also lays a foundation for establishing a more accurate prediction model of in-chamber earth pressure.(2)According to the classification method of soil excavatability proposed in this paper,after excluding the shield tunneling parameter samples of the reinforcement section at the head and tail of the tunnel,the prediction model of in-chamber earth pressure is established based on the neural network prediction method.The neural network is optimized by using particle swarm optimization(PSO)and genetic algorithm(GA),so as to find the optimal solution of neural network prediction weight and threshold as much as possible and improve the prediction accuracy of the model.At the same time,the global sensitivity analysis of the model is carried out.It is found that in addition to the screw conveyor rotation speed and advancing speed,adjusting the total thrust force can also quickly adjust the silo pressure,and the coupling effect between different excavation parameters is obvious.(3)Based on the prediction model of in-chamber earth pressure established in this paper,the optimization mechanism of tunneling parameters is proposed.After integrating the three parts of identification of soil excavatability,intelligent prediction of in-chamber earth pressure and optimization of excavation parameters proposed in this paper,an intelligent prediction and regulation system of in-chamber earth pressure is developed based on the GUI function of MATLAB platform.The human-computer interaction function is good.(4)Using the monitoring data of shield tunneling parameters and taking the limit support force obtained by numerical simulation as the expected value of in-chamber earth pressure,the optimization mechanism of tunneling parameters proposed in this paper is verified.The results show that the adjustment of tunneling parameters and the control of in-chamber earth pressure can be better realized.