Font Size: a A A

Comparative Research On Settlement Prediction Models Of Loess-Filled High Embankment

Posted on:2017-04-23Degree:MasterType:Thesis
Country:ChinaCandidate:H J LiFull Text:PDF
GTID:2272330503957166Subject:Civil engineering
Abstract/Summary:PDF Full Text Request
In recent years, China’s economic development in the central and western regions is getting better and better, the demand for high-grade highway construction is becoming more and more. The provinces like Shaanxi, Gansu, Shanxi and others which located in the Loess Plateau are typical regions. Because of the influence of specific natural geographical conditions, in road construction process often encountered mounds and gully, there appear a high filling way to build up embankment. This form of embankment is more likely to cause uneven settlement, resulting in pavement cracking and other diseases, affecting the normal use of the road. The settlement observation and settlement prediction is conducive to grasp the deformation of embankment at any time, so as to make corresponding measures to ensure the safe use of the highway. Therefore, it is necessary to do settlement observation and predict the post construction settlement of high fill embankment.According to the "high fill embankment heavy hammer tamping compaction technology in loess area and its related research about settlement deformation law" project, based on the high embankment in Yanjiafeng section of Taihang Road project in Taiyuan, Shanxi province, The layered settlement instrument was used for in situ monitoring of embankment deformation which lasted a year.Taking settlement measured data as the research object, the loess filling high embankment settlement rules was systematically analyzed, and the post construction settlement prediction model was established and contrasted. The research results show that:The two road monitoring point in one section, in the case of the basically same conditions, although there are differences in settlement, the trend of the settlement is the same and the settlement is similar. One section embankment settlement point will appear differential settlement.According to the measured data of settlement, five prediction models are established. They are the Poisson model, hyperbolic model, power model, exponential model and logarithmic model, the prediction values by five models are compared with measured values. The results show that five models all have a high degree of fitting of the short-term settlement data, but the Poisson model, logarithm model prediction value is smaller, the hyperbola, power and exponential model predicted value is larger. Comprehensive comparison of the five models of the square error and the mean absolute percentage error of two indicators, the Poisson model and the exponential model of the prediction accuracy is higher. The Poisson model has the convergence value, but the convergence speed is too fast. The other four models settlement values with time grow are the infinite and cannot predict the final settlement.Combining with construction in this paper, the Poisson model and the exponential model are selected to carry on, two kinds of model error is contrary to the trend, can better complementary advantages. Its squared error and mean absolute percentage errors were reduced and the prediction accuracy is improved, and it can provide a prediction of the final settlement of the embankment, and provide a reference for the guidance to the similar engineering.
Keywords/Search Tags:high embankment, loess, in situ monitoring, post-construction settlement, settlement prediction, combination model
PDF Full Text Request
Related items