<<
[an error occurred while processing this directive] | [an error occurred while processing this directive] >>
Vault Settlement Prediction for a Metro Cross Passage Based on Time Series
(1 College of Construction Engineering, Jilin University, Changchun 130021; 2 College of Architecture Engineering, Beijing University of Technology, Beijing 100022)
Abstract For a station on Changchun Metro Line 1, it is necessary to monitor the deformation of a cross passage during construction by the CRD method. Considering that the erection of air ducts may impede settlement monitoring, prediction of the vault settlement for a typical section becomes even more important. Therefore, an ARMA model based on time series was set up based on data preprocessing for the ingate vault settlement and the truncated characteristics of its autocorrelation function and partial autocorrelation function. Values of relevant parameters were primarily determined based on the recursion formula of the autocovariance function and the spectral density function of the MA model, and further optimization was carried out for these parameters utilizing a Newton iterative algorithm. A model was established based on the settlements of the first 100 days after cross passage excavation and was used for predicting the settlements of the next 10 days. Finally, a comparison was carried out between the results from a regression prediction and the ones from field monitoring in order to verify the feasibility of this model. The results show that the errors of the first five days' prediction results are smaller than that of the regression prediction, and they tend to be consistent with each other with the passing of time, proving the advantages of time series for short-term prediction.
Abstract：
For a station on Changchun Metro Line 1, it is necessary to monitor the deformation of a cross passage during construction by the CRD method. Considering that the erection of air ducts may impede settlement monitoring, prediction of the vault settlement for a typical section becomes even more important. Therefore, an ARMA model based on time series was set up based on data preprocessing for the ingate vault settlement and the truncated characteristics of its autocorrelation function and partial autocorrelation function. Values of relevant parameters were primarily determined based on the recursion formula of the autocovariance function and the spectral density function of the MA model, and further optimization was carried out for these parameters utilizing a Newton iterative algorithm. A model was established based on the settlements of the first 100 days after cross passage excavation and was used for predicting the settlements of the next 10 days. Finally, a comparison was carried out between the results from a regression prediction and the ones from field monitoring in order to verify the feasibility of this model. The results show that the errors of the first five days' prediction results are smaller than that of the regression prediction, and they tend to be consistent with each other with the passing of time, proving the advantages of time series for short-term prediction.
2016, Vol. 53 