Research On Prediction Measures Of Subgrade Settlement Of High Speed Rail In The Gobi Desert

ABSTRACT:Compared to the ordinary railroad, the requirements of High-Speed Railway's embankment settlement is more stringent. We must ensure that the settlement value meet the requirements before lying, therefore, it is important to accurately predict the embankment settlement and stability time. There are many ways to predict the embankment settlement. Curve Regression is used in "High-speed rail design" which requires a lot of experimental data, and certain restrictions during its using. Theoretical analysis, numerical analysis, artificial intelligence and mixed use are commonly used method in the engineering.The second high-speed railway from LanZhou to XinJiang is the first high-speed railway in the northwest desert region of China; we have never met such special geological conditions before this project, such as the unique desert soil in northwest, the soft soil which was studied from the construction of high-speed rail etc. So far, there is no data about the embankment settlement of high-speed railway based on the desert soil, the research on the soft soil is also very limited. What's more, the prediction method mentioned in the "high-speed railway design" did't includes the situation of desert soil.On the engineering background of second high-speed railway from LanZhou to XinJiang we made the following work about the subgrade settlement of the soft soil and desert soil in this paper(1)By summarizing the research situation, combined with the project of second high-speed railway from LanZhou to XinJiang, we select the specific desert soil in northwest of China and the soft soil which was studied from the construction of high-speed rail as the test section, and do some work about the settlement observation.(2) Analysis the settlement data. Three profiles of each test section is selected for analysis to ensure the accuracy, draw the curve of settlement-time, determing the two test sections have reached stability by solving the sedimentation rate(3) the hyperbolic model and exponential model are commonly used in the the method of curve fitting. Analysis the settlement data of the two test section using this two methods, find out the predicted curve, evaluate the accuracy of curve fitting and forecasting, given the applicable conditions of the two prediction methods and the regular of subgrade settlement deformation. On this basis, analysis the settlement curve furtherly, given the regular of subgrade settlement deformation more accurately(4) The gray theory was apply to analyze settlement data of this two test section, find out the gray prediction curve, observe the laws of settlement and compared with the speculation made by the previous step, further determine the settlement laws. And summarize the advantages and disadvantages of the gray theory.(5) By the way of neural network of artificial intelligence to predict the final settlement of the subgrade, analysis the results of the neural network, and point out their advantages and disadvantages, which is based on the dynamic time series prediction module and the static BP neural network.(6) By several methods of the above to find out two laws of subgrade settlement and point out some way of prediction and suitable conditions for Northwest Settlement, we can provide reference to foundation settlement in the future.