Modeling And Control Of EPB Shield Tunneling Machine

Ground deformation control has always been the technical focus and difficulty of shield con-struction in tunneling.And the most effective approach to avoid ground subsidence is to ensure earth pressure balance of the cutting surface.In order to achieve earth pressure balance,the con-struction team is supposed to control the earth pressure of working chamber accurately to balance against the front water and soil pressure to realize force equilibrium of the cutting head.There-fore,the stress field of the water and soil undergound will not change greatly and ground collapse can be avoided.This indicates that Earth Pressure Balance(EPB),as a technique of stabilizing the cutting face,can only be achieved by the existance of an automatic control system.This pa-per mainly includes two parts,modeling and control,to acquire reasonable and complete earth pressure control of shield machine.The part of modeling mainly includes the modeling of earth pressure balance control system of shield machine and the soil parameter modeling of EPB system.The former is related to de-sign structure of the controller,while the latter is helpful for the exact achievement of the control algorithm.The earth pressure balance control system model of the shield machine is mainly com-posed of hydraulic transmission system,hydraulic control components,hydraulic actuators and other models,which always contain lots of nonlinear and uncertainties characteristics.While the control methods truly based on mechanism model in the previous research seldom considered these characteristics.Therefore,in order to match the actual system more precisely,a nonlinear model of EPB control is established with consideration of particular parametric and multi-unmodeled uncertainties in this paper,which mainly involves the soil pressure dynamic,the screw conveyor-hydraulic motor dynamic and the hydraulic-system pressure dynamic model.As for the soil pa-rameter model of the EPB system,the difficulty that the front earth pressure can not be accurately measured or acquired can be solved.Therefore,a machine learning model is adopted to extract the features of the soil parameters and the established statistical regression model can predict the parameter regression with small error.The part of control mainly introduces an adaptive control algorithm based on the above two models,which can handle the particular parametric and multi-unmodeled uncertainties.A back-stepping adaptive control algorithm is proposed and possesses satisfactory control performance.However,two difficulties are derived from the algorithm:the parameter estimation is unknown and the partial derivatives are hard to be calculated in the virtual input term.Therefore,the algo-rithm is optimized and the robust adaptive control algorithms based on tuning function and com-mand filter are proposed,which solve the defects in the original control algorithm and are both theoretically proven that the state of the control system is globally uniformly ultimately bounded with adjustable transient performance,small steady tracking error and converged parameter esti-mation.