Research Of The Subgrade Settlement Prediction Model Base Of The102National Road From Changchun To Dehui

Since the1980s, the transportation infrastructure has been playing anincreasingly important role in national economic development. Thus, roadconstruction enters a period of rapid development. Up to2012, China’s total mileagehas reached2.03-2.1million km, achieving the full realization of "five vertical andseven horizontal" national trunk road.90percent of the city motorway has alreadyconnected with each other, with the total mileage of50,000km. The roadbed is animportant part of the highway projects, which good or bad performance has anextremely important role to traffic safety, road construction and operation. In thehighway construction process, roadbed settlement is the most important factoraffecting the quality of highway construction, so in the process of highwayconstruction, in order to ensure the embankment construction quality, changes of theroadbed settlement must been grasped immediately. In addition, according to changeinformation of the roadbed settlement, construction time of each structural layershould been adjusted timely, so as to avoid large embankment impact on the quality ofhighway construction. For current or future highway construction, there will be muchmore roadbeds going through the soft soil with poor foundation. Thus, settlementprediction of roadbed appears extremely more important.With highway development and the growing number of engineering software,information construction has already been used for the construction of civilengineering soft ground, high fill roadbed and other adverse geological section.Information construction is the method including the process of the constructionmonitoring, information analysis, and guiding construction or modifying design.Therefore, the use of information construction can timely find to the problem of largesettlement of the roadbed or complex geological conditions, so as to timely eliminateroadbed risks, and provide the most scientific and rational basis for the reasonablearrangement the pavement structure layer construction schedule. Settlement predictionof roadbed and the determination of settlement stability control indexes are the mostimportant content of roadbed information construction. Conventional one-dimensional consolidation theory calculation using soil test index is the commonly used roadbedsettlement calculation method, but the result is often rather different with themeasured result. Due to the roadbed settlement belonging to the three-dimensionalmodel and its actual situation very complex, using the settlement observation data toestimate post-settlement and final settlement has important practical significance.In order to establish the roadbed settlement prediction model library of the102State Road from Changchun to Dehui, this paper carried out the following fouraspects work, based on Jilin Province Transportation Research Project,“102StateRoad roadbed anti-freeze-thaw stability control and Monitoring Approach”(2009-1-25):1. Introduce the main methods and basic principles of the roadbed settlementmonitoring. Based on the actual situation of the102State Road from Changchun toDehui arterial road, select six different roadbed settlements transect forms forobservation, and the settlement point has been optimized layout, with high survivalrate and good results.2. Make comparative analysis of general prediction calculation model and othercommonly used prediction methods based on the "norm", to explore the merits anddemerits of the various models as well as their applicable scope. Use the aboveseveral predictive models to predict the actual settlement data of the observation pointB in K1144+280sections.3. Use the gray relational theory to make sensitivity analysis on the factors thataffect the roadbed settlement in the seasonally frozen area. By establishing the grayrelational data matrix of roadbed settlement in the seasonally frozen area, calculate thegray relational coefficient and the gray relational degree between the sensitive factorschange and roadbed settlement in the seasonally frozen area. Select the factor withlarge gray related degree as the input layer, and the corresponding settlement as theoutput layer, to make study on the methods and effects of settlement prediction in theseasonally frozen area by BP neural network.4. Make analysis on the prediction method by Matlab visual function to and usethis program to deal with the measured data of102national roadbed settlements to determine the effect of different models in different settlement prediction of roadbed,and finally select the most suitable roadbed structure in the seasonally frozen areafrom the six different roadbed structure layers.