基于LSSVM-ARMA地铁站基坑变形滚动预测分析

发布时间：2018-05-10 13:41

本文选题：地铁基坑变形 + LSSVM-ARMA　；参考：《河北工程大学》2017年硕士论文

【摘要】：城市规模的发展促进了城市地下工程的建设,其中以地铁的发展尤为显著,城市深基坑工程不断增加。然而施工过程中周围环境、地质条件和工艺的复杂性,使得深基坑开挖的稳定性难以保证,基坑变形的问题愈加突出,基坑变形的预测已经成为深基坑施工中必不可少的内容。针对高精度的预测基坑变形量,提出了基于小波变换的LSSVM-ARMA(最小二乘支持向量机-自回归移动平均模型)模型,实现基坑变形时间序列滚动预测。首先,由于基坑变形监测频率的不同,利用三次样条插值法对地铁站基坑变形数据进行插值。其次由于基坑变形数据具有高度的非线性,变形数据包含了土力学变化的内在趋势和一定的随机性,利用小波变换对基坑变形数据进行分解和重构,有效区分基坑变形数据的趋势项和随机项。然后,利用建立的LSSVM-ARMA模型,对基坑变形数据的趋势项和随机项进行时间序列的滚动预测,把趋势项和随机项的预测值之和作为最终的预测值。并将LSSVM-ARMA模型的预测值和PSO-LSSVM(粒子群-最小二乘支持向量机)模型以及SVM(支持向量机)模型的预测值进行对比分析,表明了LSSVM-ARMA模型的预测精度高。对广州基坑及类似工程施工安全性的提高具有重大的意义。最后,设计了基于MATLAB的GUI(图形用户界面)预测系统,实现LSSVM-ARMA模型预测过程的可视化,使预测过程以图形化的结果动态的展现出来,具有重要的实用价值。
[Abstract]:The development of city scale promotes the construction of urban underground engineering, especially the development of subway. However, because of the complexity of surrounding environment, geological conditions and technology in the construction process, it is difficult to guarantee the stability of deep foundation pit excavation, and the problem of foundation pit deformation becomes more and more prominent. The prediction of foundation pit deformation has become an essential content in deep foundation pit construction. In order to predict the deformation of foundation pit with high precision, the LSSVM-ARMA (least square support vector machine-autoregressive moving average model) model based on wavelet transform is proposed to realize rolling prediction of foundation pit deformation time series. Firstly, because of the different frequency of foundation pit deformation monitoring, cubic spline interpolation method is used to interpolate the foundation pit deformation data of subway station. Secondly, because the deformation data of foundation pit is highly nonlinear, the deformation data contain the inherent trend of soil mechanics change and some randomness, so wavelet transform is used to decompose and reconstruct the deformation data of foundation pit. The trend term and random term of foundation pit deformation data are effectively distinguished. Then, by using the established LSSVM-ARMA model, the time series rolling prediction of the trend term and the random term of foundation pit deformation data is carried out, and the sum of the predicted values of the trend term and the random term is taken as the final prediction value. The predicted values of LSSVM-ARMA model are compared with those of PSO-LSSVM (Particle Swarm Swarm least Squares support Vector Machine) model and SVM (support Vector Machine) model. The results show that the prediction accuracy of the LSSVM-ARMA model is higher than that of the PSO-LSSVM (Particle Swarm Swarm least Squares support Vector Machine) model. It is of great significance to improve the safety of foundation pit and similar projects in Guangzhou. Finally, a gui (graphical user interface) prediction system based on MATLAB is designed, which realizes the visualization of the prediction process of the LSSVM-ARMA model, and makes the prediction process display dynamically with graphical results, which has important practical value.
【学位授予单位】：河北工程大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TU753

【相似文献】