The anti-floating design of different stratum structures varies greatly. Anti-floating problems of subway stations in water-rich stratum should arouse people’s attention. The anti-floating failure mode of the subway station in a water-rich stratum is obtained by combining the indoor model test with ABAQUS finite element simulation, and specific anti-floating measures are suggested based on the findings. The typical Xiamen urban rail transit station is used as the research object. The results show that the overall anti-floating failure mechanism of the subway station situated in a water-rich stratum is caused by friction on the sidewall of the station. The floating displacement growth rate is accelerating, yet the structure is not suddenly floating. The strain at the floor and column joint is larger than other parts of the floor. The trend of pore water pressure at the soil under the bottom floor is consistent with the water pressure after the station structure floats. The pore water pressure at the soil under the bottom floor needs to be reduced, and the reduction coefficient is about 0.79-0.85. In the anti-floating design, it should be considered to set up retaining structures such as continuous underground walls to change the seepage path and increase the head loss.

In order to obtain a similar model of deeply buried limestone with rock burst propensity, the rock burst similarity criterion is established based on the principle of rock burst energy based on the magnitude analysis method, and the relevant parameters of the rough rock are estimated by referring to the relevant parameters of the proto rock and the similarity criterion of the rock burst. The rock burst propensity index, elastic properties and dissipation energy were introduced, and the orthogonal test scheme of 4 factors and 5 levels as adopted based on the quartz powder content, barite powder content, gypsum cement mass ratio and water content, and the uniaxial compressive test was carried out to obtain various parameters. The sensitivity analysis of the test results was carried out and the preferred specimens were carried out in the secondary optimization test to improve the proportion, and the parameter error of the similar model between the specimen and the target was reduced to less than 10%, and the results showed that the error between the similar model and the target similar model after the second optimization was 5%. Moreover, dynamic impact mechanical tests are carried out on optimized specimens with different impact velocities, and the peak stress law accords with the law of limestone dynamic failure, which provided a reference for the indoor study of statics and dynamics in limestone interior.