Analysis and prediction of nonuniform deformation in composite strata during tunnel excavation

Due to the influence of composite strata, traditional monitoring and measurement cannot accurately reflect the nonuniform deformation and corresponding local damage to tunnels. However, the nonuniform deformation precise analysis and effective prediction of composite strata tunnels are indispensable in related projects. This analysis established a nonuniform deformation classification method and 135 groups of numerical models for a composite strata tunnel. The nonuniform deformation classification method involves four reference-defined in-dexes, the average deformation coefficient, variable deformation coefficient, segmental deformation coefficient, and abnormal deformation coefficient. The numerical model took the dip angle of the formation, formation proportion, in-situ stress, and pore pressure as composite strata model conditions, with four reference indexes in the nonuniform deformation classification method as the output parameters. The intelligent model of the least squares supports vector machine (PSO-LSSVM) was innovatively utilized to predict nonuniform deformation in composite strata during tunnel excavation. The prediction results of the PSO-LSSVM are consistent with actual engineering data relying on the engineering situation of Teermo Tunnel. This approach can provide targeted guidance for a composite stratum tunnel engineering evaluation system.