Intelligent Back Analysis Method for Surface Settlement during Excavation of Double Track Tunnel Based on Improved GA-PSO
SHEN Qingmeng
YANG Yukun
LI Yue
LI Tao
WANG Xiaolong
LI Feng
Abstract：To address the value accuracy of ΔA1,ΔA2(the sectional convergence areas of the preceding and subsequent tunnels)and tanβ1,tanβ2(the tangent values of the main influence angles of the preceding and subsequent tunnels)in the study of double track tunnel settlement under random media theory,an intelligent displacement inverse analysis method was proposed based on an improved genetic algorithm-particle swarm optimization(GA-PSO).The method used Latin hypercube sampling to obtain initial samples in a multi-dimensional parameter space.Real-valued encoding and a monarch selection crossover strategy were applied to improve the standard GA.PSO was then employed to further perform global and local searches in the high-quality individuals which selected by the improved GA,ensuring an efficient search for optimal parameters.In the inverse analysis of the section example between Xibahe station and Sanyuanqiao station of Beijing metro line 12,the calculated values were compared with the measured values.Suitable parameter values were obtained,and tunnel settlement for subsequent excavation was predicted.The results show that the double track tunnel multi-parameters model is better aligns with engineering practice.By comparing the calculated settlement values with the measured settlement values,it is found that the relative error ranges from 1.36％to 12.28％.At the same time,the improved GA-PSO hybrid algorithm show more accuracy compared to the individual improved GA and PSO algorithms.The research results expand the applicability of random medium theory and enriches the method for determining parameter values in double track tunnel random medium theory.
Keywords：double track tunnelintelligent algorithmsrandom medium theorydisplacement back analysis
Publication Date：2024-12-25
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：12( 69-80 )