结合三次样条和时序模型的桥墩沉降预测

Bridge pier subsidence prediction based on cubic spline and time-series model

在时间序列模型中,等时间序列模型建模过程简单,预报准确,但观测时间是不等间隔的;三次样条插值可以对一定的区间进行有效内插,得到等时间间隔的拟合数据。针对两种模型的优点,该文结合等时间序列模型和三次样条插值进行高铁桥墩的沉降预测。首先通过三次样条插值将数据进行等时间间隔处理,然后利用得到的数据建立时间序列模型,最后通过三次样条插值求得相应观测时间的预测值。以某高铁桥墩沉降的观测点为例,分别进行原始数据时间序列建模以及三次样条插值和等时间序列组合模型建模。实验结果证明,组合模型的预测精度更高。

In the time-series models,time series model has the advantage of simple modeling and good forecast accuracy,but its observation time is unequal;cubic spline interpolation can obtain fitting data with equal time intervals by interpolation within a certain range.Combining the advantages of two models,a new model was proposed for pier settlement prediction.Firstly,cubic spline interpolation was used for equal time interval processing of data;secondly,the obtained data was used to establish the time series model;finally,the prediction value in corresponding observation time was obtained by cubic spline interpolation.Taking high-speed rail piers settlement observation points as an example,original data time-series modeling and the proposed combination modeling was compared.Experimental results showed that the proposed method has higher prediction precision.