Intelligent prediction model of tunnelling-induced building deformation based on genetic programming and its application

Abstract

This paper aims to explore the ability of genetic programming (GP) to achieve the intelligent prediction of tunnelling-induced building deformation considering the multifactor impact. A total of 1099 groups of data obtained from 22 geotechnical centrifuge tests are used for model development and analysis using GP. Tunnel volume loss, building eccentricity, soil density, building transverse width, building shear stiffness and building load are selected as the inputs, and shear distortion is selected as the output. Results suggest that the proposed intelligent prediction model is capable of providing a reasonable and accurate prediction of framed building shear distortion due to tunnel construction with realistic conditions, highlighting the important roles of shear stiffness of framed buildings and the pressure beneath the foundation on structural deformation. It has been proven that the proposed model is efficient and feasible to analyze relevant engineering problems by parametric analysis and comparative analysis. The findings demonstrate the great potential of GP approaches in predicting building distortion caused by tunnelling. The proposed equation can be used for the quick and intelligent prediction of tunnelling induced building deformation, providing valuable guidance for the practical design and risk assessment of urban tunnel construction projects.