Research On Stratified Horizontal Displacement Regular Pattern Of The Special And High-fill Embankment Roadbed

With the deepening of China's reform and opening up,transportation industry gets rapid development.The express highway plays a very important role in society and economy development.The scientific workers are puzzled by the road quality,for example,lateral deformation of subgrade during construction and operation.The forecast on lateral displacement of special road,such as soft soil,high cut and fill subgrade,is very important.Choosing the proper deformation forecast model is the important work of laws research.At present,the study is not sufficient.It goes without saying that it is necessity to seek proper forecast models for lateral displacement.Being based on the deformation mechanism of lateral displacement,the paper studies the position of maximal lateral displacement.On the basis of comprehensive analysis,it researches the results of multiple prediction models.And then it builds the deformation forecast model,and analyses the characteristic and boundedness of these models.According to the landform and geology,it models the laws of deep horizontal displacement by using the finite element analysis software-MIDAS/GTS.The results coincided with the actual measured value.This paper focuses on the research of gray model GM(1,1)and logarithmic curve-fitting,and builds the forecast model of deformation on different measuring points and depths.According to the precision classes standards,the precision of gray model GM(1,1)and logarithmic curve-fitting is higher than others.The approximation rate to measured values is faster than logarithmic curve-fitting.Pearl curve reaches prematurely the mature stage,so it predicts poorly in the latter part.The exponential smoothing method depends on the measured values so strongly that it is not suitable for middle and long period prediction.In all,the paper puts forward a trend and correction method being named trend correction model,which is combined trend value with correction value.After we modify trend value,the correction model tallies better with the measured value,and has a higher fitting degree.Its precision is obviously better than trend value,and it can reflect actual changes more objectively.So this model is feasible,and has a certain reference value and practical significance.