Mingshan Tunnel Construction Period Settlement Prediction Based on DE-SVM

 

Zhongle Lu *

PhD student, Faculty of Engineering, China University of Geosciences (Wuhan)
*Submitting author; e-mail: lzlcug@foxmail.com

Li Wu**

Professor, Faculty of Engineering, China University of Geosciences (Wuhan)
**Corresponding author; e-mail: lwu@cug.edu.cn

Xuewen Zhang

ME student, Faculty of Engineering, China University of Geosciences (Wuhan)
e-mail: hover.aoxiang@gmail.com

Ruifeng Zhou

ME student, Faculty of Engineering, China University of Geosciences (Wuhan)
e-mail: zrfcly1215@gmail.com

 

 ABSTRACT

In tunnel construction period, there exists a complex, nonlinear relation between time and settlement, as a research branch of tunnel time-space effects. The paper proposes the combination of differential evolution and support vector machine to form the DE-SVM model applied to accurately predict tunnel arc top settlement based on the site survey. Through introduced and analysis of the SVM function and its system structure, and DE optimal process, the DE-SVM can be suitably applied to tunnel settlement prediction to achieve ideal effects. By compare on the same sample data regressions by DE-SVM and GA-SVM, the error of prediction by DE-SVM is obviously less than that of the other model. From the case of Mingshan High Speed Railway Tunnel, the advantages of DE-SVM are expounded that it owns the characters of higher accuracy, faster convergence, and stronger adjustability; therefore DE-SVM settlement prediction model can be widely used to the similar construction required the high precision and the simple approach.

Keywords: High speed railway tunnel; Settlement prediction, Differential evolution (DE); Genetic algorithm (GA); Support vector machine (SVM).

Get the entire paper (pdf)    Go back to the TOC