Fill Roadbed Settlement Prediction And Its Additional Stress Analysis In Season Frozen Area

Subgrade subsidence in season frozen area affected by the temperature field, moisturefield and stress field at the same time, in the process of its deformation affected by manyfactors, mainly from the season frozen area geological conditions, height of subgrade filling,filling materials, construction technology and the comprehensive influence of climateconditions etc. Subgrade under the effect of various factors affecting joint will cause unevensettlement. The uneven roadbed deformation caused by uneven settlement and seasonalfreezing and melting has become the main damage reason of the pavement structure. Due toinhomogeneous deformation of roadbed will generate additional stress of pavement structurelayer, superimposed stress produced by the action of vehicle load will make the pavementcracks, easy to cause pavement early damage.The paper was based on the science and technology project come from communicationsdepartment of Jilin province “Monitoring methods study of Freeze-thaw stability of thenational highway102line” as the backing, with the study of Beijing-Harbin line first classhighway in Changchun to Dehui period of test road as the research object, analysised theroadbed settlement prediction and its additional stress in the season frozen area, through theresearch of this paper has the following findings:1Based on the comprehensive analysis of roadbed subsidence monitoring technologyand the status of forecast methods, according to the practical engineering Beijing-Haerbinline first class highway roadbed itself characteristics and with the combination of existinghighway roadbed subsidence monitoring situation established the roadbed settlementmonitoring plan, and monitor the roadbed settlement monitoring point was set. All testsections of the measured data combining with the subgrade settlement mechanism and theroadbed settlement law of development are analyzed in detail.2Summarized the forecast method based on the measured data, and introduced the prediction method of static and dynamic prediction method. According to the test section ofthe site measured data using the hyperbolic method, exponential curve method and greytheory to the real-time prediction research, with a study found that application of hyperbolaand exponential curve, GM (1,1) grey theory prediction model results with the fieldmeasured results is better, shows that the three methods in roadbed settlement prediction hascertain applicability. Compared the three, hyperbola and exponential curve predictionmethod is convenient and simple, the prediction of final subgrade settlement calculationprocedure is simple and easy to understand, and the memory footprint is small. But thismethod needs to meet certain conditions, or prediction accuracy will fell dramatically. Greytheory prediction model is established on the basis of strict mathematical theory, and theprediction results with higher precision. Grey theory forecast of the parameter is fixed, itwould not affected by geological conditions and external environment factors, but as the dataincreased the corresponding calculation also increased, thus reduced the practicability of thismethod.3Based on the transverse longitudinal settlement according to the characteristic of spacecurved surface changes. Proposed settlement of roadbed transverse law according to theparabola change, lateral each measuring point number rule changes over time according tothe hyperbolic or index to study the settlement of road cross section are studied. This methodis based on has been observed settlement data, the least square method is used to determinethe settlement of roadbed lateral surface, forecast the final settlement. This method cancomprehensive transverse impact factors, determine the maximum settlement of section, andto study the transverse structure of uneven settlement rule.4This paper presented index curve and neural network hybrid modeling method to predictroadbed settlement. According to the method that roadbed settlement rules are divided intoascertain and uncertainty, the ascertain part using the exponential curve method modelingand uncertainty using neural network modeling, and mixed sedimentation model is obtained.By using the index method’s characteristics of simple and quick, reflects the main law ofsubgrade settlement, at the same time, and use the neural network to predict the uncertainparts of roadbed settlement, due to the neural network only to forecast the settlement of uncertainty, to achieve the same accuracy, the calculation scale of hybrid neural networkmodel in the modeling is smaller to the single neural network model, the training speed isfast. Through the prediction of section of test section of the measured data, proved that takethe two methods together, the forecasting result will be more accurate and more efficient.Provides a new effective method for roadbed settlement prediction.5For the influence factors of subgrade differential settlement and settlement rule,summarizes two roadbed settlement patterns in season frozen area. And use the Winklersubgrade model to analyzed the additional stress caused by roadbed settlement, and theanalysis results show that the maximum subsidence influenced the stress obviously, andsubgrade coefficient value’s influence on the lateral stress is lesser, it can’t reduce theadditional stress by increasing the subgrade coefficient. Subgrade modulus and stress has alinear relationship, this shows that the subgrade modulus E higher the greater requirement ofroadbed. Using three-dimensional finite element method to analyzed the subgradedifferential settlement of the two kinds of mode, the results of the analysis and the Winklersubgrade beam analysis results are basically identical. Proved that the validity of the Winklersubgrade beam analysis of uneven settlement.