Subgrade Settlement Law Affect Factor Analysis And Settlement Prediction

Due to the rapidality and high technology standard of high speed railway and the strict requirement of its subgrade settlement, controlling the deformation of passenger railway line embankment has been the most obvious characteristic. Since constructing Qinghuangdao-Shenyang Passenger Railway Line in 1999, the construction of passenger railway line has been roundly achieving its peek time in China. Wuhan-Guangzhou Passenger Railway Line is the first line in which ballastless track is adoppted in the whole line, so the research on the developing law and the embankment mechanism of the settlement, and the analysis that the influencing factors of embankment settlement will make effect on subgrade settlement has a great significance in guaranteeing the passenger railway line's operational safety and its economic efficiency.The paper taking the Comprehensive Experimental Section of Wuhan-Guangzhou Passenger Railway Line as the project background, a systematic and complete site monitor has been done and measures including theoretical analysis, nonlinear multivariable regression mathematical statistical analysis, BP artificial neural network using fortran programme language have been adopted to make a comprehensive, in-depth and meticulous research for settlement characteristic of the passenger railway line embankment on the basis of classifying, generalizing and analyzing a large amount of the field measureed data. Its main tasks are the following:(1) The scene observation projects were identified and programme components were planted. After that, the field measurement data were gained through nearly one year, and then systematic collation and analysis of the testing result were done.(2) According to the field measurement data, a research on the developing process and law of settlement was done and some meaningful conclusions to engineering were drawn. (3) A multi-linear regression equation of the relationship were transformed and evaluated and then the likely causes were analyzed through analyzing the field measured data and a variety of factors on subgrade settlement, using multivariate statistical methods to seek an intrinsic relationship between the final settlement and the effective factors on subgrade settlement.(4) On the foundation of layer-summation, a multi-nonlinear regression equation of settlement deformation and the height of construction, elastic moduli, thickness of the compressed layer was gained, and then transformed into a multi-linear regression equation, and the effective law and degree of every factors towards settlement were studied according to the field measured data.(5) Multi-nonlinear regression equation about multi-nonlinear regression equation of settlement deformation and the height of construction, elastic moduli, thickness of the compressed layer, the time were established. Methods of multiple non-linear regression analysis to the time series data were given. Through the application of cases and the usage of the field measured data, existing problems were analyzed, and their applicability was proved, and then the direction for future research was figured out.(6) A predictive model were built on the foundation of artificial neural network, combining the time series data with mathematical statistical analysis of network section data. Fortran procedures in BP artificial neural network were compiled and applied in this project. A satisfying outcome were reached through selecting joint samples under various models of enbankment to have neural network training and forecasting the settlement of every section respectively.