Settlement Prediction Of Highway Soft Base Based On Nonideal Observation Data

During the construction of the freeway,the work of safety monitor of the roadbed sedimentation provides frist-hand information for all kinds of researches,with large numbers of local data.As the work of sedimentation monitor on freeway belongs to the field of the safety monitor,the dependability of the data affects related analysis results,and impacts the engineering quality directly.In order to reduce the adverse impacts of inevitable monitoring system error and artificial factors on data analysis ,I advise to pretreatment the data.In order to reflect the fact of roadbed settlement,I introduced a weight changeable combination forecast model in soft base settlement forecast to furthest advance the veracity of soft-base settlement forecast.The main work of this dissertation is listed as follows:1.On the basis of error analysis for settlemeng monitoring of advanced highway soft-base,analyzed the cause of abnormal data.Point out the treatment for abnormal data which caused differet reasons.I introduce one variant linear regression analysis and grey relational analysis to analysis abnormal data.2.Apply hyperbola model,Pearl model and GM(1,1) model into the interpolation of soft-base settlement data,through calculation to compare the interpolation effective of three models.Discussed the revision method of the model in order to improve the precision of settlement data interpolation.3.Summarized the models available and pointed out the disadvantages of the models. In this paper, through settlement prediction of the actual case of Huai-Yan expressway, some main factors in the hyperbolic fitting method and Pearl model are studied. Some measures to improve predicting accuracy are proposed.4.Aimed at the problems of high restriction and low forecast veracity in subside analysis of single model,I induced combination forecast model in this dissertation.Applied the weight changeable combination forecast model to forecast soft-base settlement and compared the result with single models in order to evaluate the weight changeable combination forecast model's applicability in the field of soft-base settlement forecast.