Learning Research On Traffic Reliability Of Qinghai-Tibet Railway Based On Statistical Learning

The Qinghai-Tibet Railway is the main form of linking Tibet,Qinghai and the mainland with modern transportation facilities.The geological conditions of the subgrade engineering in the permafrost area of the Qinghai-Tibet Railway are extremely unstable,and subgrade settlement will inevitably occur.The subgrade settlement will affect the mechanical properties of the system,deteriorate the long-term service performance of the track structure,damage the ride comfort,and even threaten the driving safety in severe cases.In view of this,this paper is driven by the field monitoring data of the permafrost subgrade elevation between the Tanggula South and Amdo section of the Qinghai-Tibet Railway,and uses statistical methods to study the long-term evolution law prediction of subgrade settlement,and the subgrade settlement and subgrade lateral differential settlement under the coupling effect of temperature and vibration.,Reliability assessment of longitudinal uneven settlement of subgrade,in order to provide certain theoretical basis and guidance for the operation and maintenance of Qinghai-Tibet Railway.First,in order to analyze the irregular deformation of the subgrade elevation of the Qinghai-Tibet Railway,a subgrade settlement prediction model based on grey neural network is presented.The residuals of GM(1,1)fitting data are used to train the BP neural network to obtain the residual sequence after training,and then calculate the new predicted value of subgrade settlement.The research results show that compared with the traditional gray prediction,the gray BP neural network model has a smaller average relative error and higher precision in predicting the settlement of the permafrost subgrade of the Qinghai-Tibet Railway,which can effectively predict the settlement of the frozen subgrade of the railway.According to the prediction results,the predicted subsidence risk points after three years and ten years were summarized.Furthermore,different from the qualitative exploration,the influence of the coupling effect of ambient temperature and driving vibration on the settlement and deformation of the subgrade and the calculation of the dynamic reliability of the subgrade are quantitatively studied.Taking ambient temperature,driving vibration stress and time as input,and subgrade settlement as output,a multiple linear regression model of subgrade settlement and deformation is established.Based on the statistical analysis of the predicted value of the regression model,the reliability evaluation method of the Qinghai-Tibet Railway subgrade is given.The goodness of fit of the model is significant,and the predicted value of the safety and reliability of the Qinghai-Tibet Railway in the future is given.The environmental temperature causes the subgrade frost heave and thaw phenomenon,and the long-term dynamic load and the accumulated deformation under the action of time cause the subgrade subsidence,all of which weaken the subgrade reliability.Then,in order to explore the lateral differential settlement of subgrade in the permafrost area of Qinghai-Tibet Railway,the comprehensive reliability assessment of subgrade yin and yang slope structure in the field environment was studied by the method of predicting and evaluating the reliability of subgrade settlement with multiple degradation factors based on Copula function.question.The change of the subgrade settlement on the shady and sunny slopes is selected as the key degradation factor of the roadbed measurement point,and the degradation model is carried out on the actual collected data according to the different degradation trajectories of the unilateral road shoulder,and the marginal distribution function of the subgrade settlement value is obtained.Then the degradation factor is established by the Copula function.The edge distribution of subgrade settlement is fused to obtain its joint distribution function,and the comprehensive reliability of subgrade measuring points is calculated.The validity of the model is verified by taking the subgrade elevation of a measuring point along the Qinghai-Tibet Railway as the engineering background.The results show that: using the constructed Copula model to analyze the settlement values of the shady and sunny slope shoulders can realize the sensitive capture of the comprehensive reliability changes of the subgrade measuring points under the time effect,and use the analysis results to formulate targeted subgrade maintenance measures.On this basis,a comprehensive evaluation of the reliability of running stability was carried out by using the national standards for the control of subgrade settlement of high-speed railways,and taking the length of the subgrade longitudinally uneven settlement area,the settlement amplitude,and the train speed as the interference factors.Using the calculation and derivation of probability distribution,the precise expressions of the stability and reliability of single-factor interference and double-factor linkage interference of subgrade settlement parameters(subgrade settlement area length,settlement amplitude)are obtained.The Monte Carlo simulation model is used to obtain the subgrade settlement area length.-Algorithm steps of the stability reliability of the settlement amplitude-vehicle speed three-factor interference.The research is helpful for early warning of high-speed train operation risks.In order to ensure the stability and reliability of the train to be above 0.95 and make the most robust decision,the m settlement area should be controlled to have an amplitude less than 9.5 mm;the settlement amplitude corresponding to the m settlement area It is recommended to control it below 12.8 mm;the amplitude corresponding to the uneven settlement of the subgrade with a large settlement area should not exceed 19.1 mm.