Study on the Construction of Tunnel Face Safety Factor Prediction Formula Based on MARS

To address the challenge of assessing the stability of rock tunnel face,a fast and efficient assessment model was proposed.A comprehensive dataset comprising 818 set of data encompassing crucial parameters like GSI,σ_c and m_i were meticulously collected from the literature.The correlation coefficients between σ_c and m_i as well as GSI and σ_c,were subsequently determined.The Latin hypercube sampling and analytical solutions were combined to construct a robust stability database specifically tailored for Grade V rock tunnel faces.Utilizing the Multivariate Adaptive Regression Splines (MARS) algorithm,a predictive formula for the safety factor of rock tunnel face was developed,and rigorously validated using on-site construction data.Comparative analysis indicates that the geometric parameters of the tunnel exert a pronounced impact on the stability than rock properties.The model exhibits good interpretability of the intricate relationships of the tunnel face safety factor to various factors,enabling rapid,accurate,and reliable computation.Consequently,the assessment model stands as a convenient on-site tool,offering expeditious evaluations of the rock tunnel face stability.