Application of Improved TOPSIS in Back Analysis of Soil Stiffness Parameters

［Objective］ When the back analysis method is used to determine soil parameters, the ideal solution obtained from field measurements is generally an intermediate value, which cannot be calculated using the traditional TOPSIS (technique for order preference by similarity to solution). It is necessary to improve the traditional TOPSIS and combine it with the entropy weight method, numerical analysis method and orthogonal test method for back analysis of soil stiffness parameters. ［Method］ The improved traditional TOPSIS is briefly described and combined with entropy weight method, numerical analysis and orthogonal test method, then applied to the back analysis of stiffness parameters in water-rich sandy stratum. 25 numerical analysis models of the schemes are established with the orthogonal test method. Four indicators including maximum ground settlement, maximum arch crown settlement, maximum arch bottom uplift and maximum clearance convergence are selected for comprehensive decision-making, and the proportional relationship and size of the stiffness parameters of the HS (hardening soil) model suitable for water-rich sandy stratum are determined. ［Result & Conclusion］ The improved TOPSIS can effectively combine numerical analysis and orthogonal experiment method for back analysis research, and make multi-index decisions on orthogonal experiment schemes scientifically, providing a new decision-making method for the back analysis theory of complex geotechnical engineering. The proportional relations between the stiffness parameters of the HS model in water-rich sand stratum are: tangent modulus of consolidation test Eref,oed=Es(0.1~0.2) (0.1~0.2 MPa compression modulus under ambient pressure), the cut line modulus of three-axis consolidation drainage shear test Eref50=1.4Eref,oed , the unloading and reloading modulus of three-axis consolidation drainage unloading and reloading test Eref,ur=2.0Eref,50 , the stiffness stress correlation index equals 0.6.