Intelligent Inversion of Mechanical Parameters of Surrounding Rock of Underground Water-Sealed Storage Cavern During Construction

Reasonably and accurately determining the mechanical parameters of surrounding rock is crucial for the stability analysis of surrounding rock and the design of supporting structures during the construction of underground water-sealed storage caverns. Based on the engineering geological characteristics of underground water-sealed storage caverns, an intelligent inversion method for the mechanical parameters of surrounding rock during construction is proposed for the equivalent continuous medium model. First, the constitutive model of the cavern rock mass is determined via laboratory and numerical tests. Then, the mechanical parameters of the constitutive model are estimated using the Hoek-Brown strength criterion and other methods, and the sensitivity of each parameter to the deformation and relaxation of the surrounding rock is analyzed. Based on these analyses, an orthogonal test scheme for the parameters to be inverted is designed. Finally, an intelligent inversion model for the mechanical parameters of the surrounding rock is constructed using samples obtained from numerical simulations and an evolutionary neural network model. Comprehensively estimating the initial value ranges of the mechanical parameters of the surrounding rock for different classes of rock mass basic quality can avoid large deviations in inversion results caused by the wide optimization range of intelligent algorithms. Sensitivity analysis of the mechanical parameters to the deformation and relaxation of the surrounding rock can reduce the number of inversion parameters. The proposed intelligent inversion method is applied to a typical underground water-sealed storage cavern project in Shandong Province. Results show that the relative errors of the settlement of the peripheral rock vault, the maximum displacement of the cavern circumference, the internal deformation, and the average depth of relaxation are all less than 10%, meeting the precision requirements of engineering applications. Therefore, the proposed intelligent inversion method exhibits high feasibility and reliability through practical application and can serve as a reference for similar projects.