Research On Dynamic Characterristics Of High-speed Railway Subgrade And Planning Key Parameters Of Rolling

The compaction quality of the High-speed railway subgrade in the filling stage and the stability and safety of the operation stage are very critical.At present,there have been some researches on the continuous compaction control technology and the dynamic response of the subgrade under the action of dynamic load,but the continuous compaction control technology still has shortcomings.For example,the planning of the subsequent rolling conditions mainly depends on the long-term construction experience,most of the research on the dynamic characteristics of the subgrade is aimed at the interior of the subgrade.This paper takes Beichen Railway Station and Gu’andong Railway Station as the background,through the continuous compaction test and in-situ vibration test of the subgrade,combined with the LSTM neural network and Abaqus numerical analysis as means:to explore the correlation between the continuous compaction detection indicator and conventional detection indicators during the rolling process of fillers and the dynamic response of the surface of the filler and the vibration wheel,and combined with the neural network to make a well-founded plan for the subsequent rolling conditions;to discuss the dynamic response and characteristics of the surface interior of subgrade under the action of dynamic load and with different elastic moduli combined with numerical analysis software.The main research conclusions of this paper are as follows:（1）The coefficients of determination between the continuous compaction detection indicator CMV and the detection indicators K₃₀ and E_vd are all greater than 0.5,and the CMV increases with the number of rolling passes;The fundamental wave amplitude of the Fourier spectrum,as well as the acceleration peak value,the HHT spectrum amplitude and the marginal spectrum amplitude of the vibration wheel,will increase with the number of rolling passes;（2）The structure of predicting the compaction energy indicator CEV is constructed through the LSTM neural network.The predicted mean absolute value percentage error is controlled by 10%,the root mean square error is controlled by 2.5,and a planning system for key parameters of high-speed railway subgrade rolling is created,which can realize the planning of rolling conditions based on this structure;（3）Under the excitation of different loads,the acceleration peak value of the subgrade surface decreases with the increase of the horizontal distance,and most of them attenuate between 0m-0.3m.The amplitudes of the HHT spectrum and the marginal spectrum increase with the increase of frequency,and the HHT spectrum gradually evolves from"single peak"to"triple peak".As for the dynamic stress inside the subgrade:the evolution law of the dynamic stress at a distance of 0m from the excitation center is not consistent at 1.2m,and it is concluded that the frequency with the more significant dynamic influence on the subgrade surface is 20Hz through the frequency sweep test.（4）The dynamic stress of the subgrade surface and its internal acceleration response characteristics under different loading displacements and frequencies were studied by ABAQUS software.It is concluded that the dynamic stress of the subgrade surface gradually changes from increasing to first increasing and then decreasing as the loading displacement increases,and the loading displacement has a more significant effect on the dynamic stress of the subgrade surface;the acceleration peak value inside the subgrade increases with the increase of loading frequency and displacement;the dynamic stress on the subgrade surface has a small amplitude with the increase of the elastic modulus of the subgrade soil,and the internal acceleration peak gradually changes from an increase with the increase of the elastic modulus to a law of first increase and then decrease.