Research Of Feasibility Of Prediction Methods For Subgrade Settlement On Beijing-Shanghai Express Railway

The observation and assessment of the settlement are the linchpins to insure the quality of track-laying of the ballastless high speed railway. The smoothness, stability and the operational security of the railway in post-construction period also rely on them, while the prediction of the settlement is the core of assessment. Nevertheless whether the widely-used curve fitting method in highway construction is suitable for the circumstance of soft ground subgrade in high speed railway construction remains a problem. So the analyzing and optimizing the curve fitting method in different working conditions has value for engineering application and academic studying, so does the researching of the features of settlement-prediction method and getting the reasonable engineer parameters.Based on the researching project "assessment of the subgrade settlement in construction of Beijing-Shanghai express railway", this thesis takes the data from the key segment DK511+450.00～DK511+807.50of the third bid section of Beijing-Shanghai express railway as researching object, which has a adverse geological condition. The feasibility of different curve fitting methods has been studied, and then, the optimal and suboptimal methods for settlement-prediction have been chosen, and their model features have been studied. Finally the reasonable engineer parameters have been chosen to guide the assessment of the subgrade settlement in construction of Beijing-Shanghai express railway.The main work of the thesis is as following:(1) The methods of observing the subgrade settlement have been systematicly introduced, so do the control and assessment gauge of subgrade settlement. The mechanism and characteristic of subgrade settlement have been briefly illustrated.(2) Hyperbolic curve method, exponential curve method, Asaoka method and grey theory have been used in prediction of subgrade settlement of the object railway section. Their feasibilities have been discussed. The optimal and suboptimal models have been chosen and mutually compared. My study shows that hyperbolic curve method and Asaoka method have the best prediction effect, with larger correlation coefficient, smaller residual and rela-tive error range.(3) The characteristics of hyperbolic curve model and Asaoka model have been studied with the aim at analyzing the influence of starting point and the time interval upon the correlation coefficient and relative error. The best prediction parameters have been chosen by checking the prediction accuracy in the long term. The method to determine starting point and the time interval of samples under the two-level load has been given. This thesis also clarifies how to determine the equal time interval in Asaoka method.