Study On Intelligent Identification Of Differential Settlement Of Metro Tunnel Based On Vehicle-Track Vibration

Due to the influence of soil properties,surrounding construction,long-term dynamic loads of trains and other factors,the differential settlement of metro tunnel often occurs,which leads to speed limit of trains and aggravates the damage of track structure,causing significant safety hazards for metro operations.However,traditional monitoring techniques cannot meet the requirements of efficient,real-time,and automated monitoring of tunnel settlement.In the study on prediction and identification of differential settlement of tunnels using artificial intelligence methods,the early monitoring data are used to predict the late settlement of the monitoring point,and it is difficult to obtain the tunnel settlement state for the whole line.Aiming at the above problems,a method for identifying differential settlement of metro tunnels based on recurrent neural network is proposed.Firstly,based on the theory of vehicle-track coupling dynamics,a coupling dynamic model of metro train-monolithic track bed is established.The differential settlement of metro tunnel is considered to analyze the vehicle-track vibration response,and the vehicle-track vibration response sensitive to the differential settlement is obtained.Further,a sample database of "settlement valuesensitive vibration index" time series is established,and then an intelligent identification model based on recurrent neural network is established,which is used to identify differential settlement of metro tunnels.The main study results and innovations of this paper are as follows:(1)A coupled dynamic model of metro train-monolithic track bed is established,taking into account the effects of longitudinal stiffness reduction of tunnels,random track irregularities,and random factors such as train mass and component parameters.Based on the theory of vehicle-track coupling dynamics,the coupled dynamic equations of the train and track structure are derived and solved,and further calculations are performed using MATLAB programming to obtain the vehicle-track dynamic response of the metro train-monolithic track bed model.(2)Based on the metro train-monolithic track bed model,the influence of tunnel settlement wavelength,tunnel settlement amplitude and train speed on the dynamic response of vehicle-track coupling system is analyzed by introducing the differential settlement condition of the tunnels.The dynamic response indicators sensitive and regular to the non-uniform settlement of the tunnel are found,and the sample database of "settlement value-sensitive vibration response" is established based on the theoretical calculations.Further,based on the safety of the train operation and passenger comfort indicators,the limited value of the tunnel differential settlement is specified.(3)Several intelligent identification models based on recurrent neural network are established,the most suitable sensitive dynamic response for identifying differential settlement of tunnels is confirmed.Further,based on the sensitive dynamic response,the settlement identification under various conditions is studied,which proves the theoretical feasibility of intelligent identification models for identifying differential settlement of tunnels.The performance differences of each intelligent identification model are also discussed.(4)Based on a practical measurement database of a certain metro line in China,the intelligent identification model is verified that the intelligent identification method for differential settlement of tunnels based on vehicle-track vibration is also feasible for engineering applications.