基于多源信息融合的路堤沉降预测方法

An approach for predicting embankment settlement by integrating multi-source information

准确预测路堤沉降对于规避风险和减小成本至关重要。传统的仅基于场地勘察数据的路堤沉降预测方法的预测值常偏离监测值。提出了基于贝叶斯理论的多源信息融合方法进行路堤沉降预测,采用有限元法模拟多层土体的固结沉降,并结合高效的马尔科夫链蒙特卡洛模拟法更新土体参数得到高维后验分布。以新南威尔士州的Ballina地区试验路堤数据为例说明了所提方法的有效性。结果表明,基于贝叶斯理论的多源信息融合方法可以有效融合勘察和监测数据,通过多源数据融合能够较准确地预测路堤沉降。对于Ballina路堤,总体上随着监测数据量的增加,路堤沉降预测预测精度逐渐提高。使用0～116 d监测数据可以准确地预测地表沉降;基于0～496 d监测数据可同时准确预测所有监测点的沉降。对于Ballina路堤,先验信息对沉降预测具有一定影响,但观测误差对预测准度影响微弱。

Accurate prediction of embankment settlement is critical for risk mitigation and cost reduction in embankment projects. Traditionally, the prediction only using data from site investigation usually deviates from the monitored settlement. In this article, it is proposed to integrate multi-source information based on Bayesian theory to predict embankment settlement. The finite element method is adopted to simulate the consolidation process of multiple soil layers, and the posterior high-dimensional distributions of soil parameters are obtained by efficient Markov Chain Monte Carlo simulation. The proposed method is validated by site data from a trial embankment constructed at Ballina, New South Wales, Australia. The results indicate that the proposed multi-sources information integration method based on Bayesian theory can effectively integrate date from site investigation and field monitoring, based on which the embankment settlement can be accurately predicted. For the trial embankment at Ballina, the accuracy of prediction is improved in terms of the overall trend as more monitored data is incorporated into Bayesian updating. The accuracy of surface prediction can be satisfied based on data from 0-116 d, while the data of 0-496 d can be used to monitor settlement for all the monitoring points. For the Ballina embankment, the prior information affects slightly on the posterior prediction, while the measurement error barely affects the prediction.