Abstract: The honeycomb structures have emerged as promising anti-collision energy-absorbing core materials due to their advantages of lightweight and having high energy absorption capacity. Its application on bridge piers can reduce the impact risk of collisions from barges. Shear-reinforced honeycombs can improve the anisotropic compressive performance of traditional honeycombs and help mitigate structural damage， yet they have received limited attention. Moreover， traditional design methods often rely on repetitive finite element simulations， resulting in low computational efficiency. To address this， this paper proposed a plateau stress prediction model based on an improved Latin Hypercube Sampling （LHS） method. First， the superior performance of the shear-reinforced origami honeycomb under oblique impact was confirmed through finite element simulations. Second， the LHS method was employed to perform uniform sampling in the multi-parameter space to obtain high-fidelity sample data in conjunction with numerical simulations. Subsequently， a nonlinear plateau stress prediction model incorporating key parameters such as wall thickness， side length， and offset distance was constructed using physics-inspired functions. Finally， the model’s accuracy was validated using random samples. The results indicated that the coefficient of determination （R²） of the predictive model reached 0.997， with extrapolation errors controlled within 10%. By integrating the LHS method with surrogate modeling techniques， this study achieved rapid and accurate prediction of the plateau stress for shear-reinforced origami honeycombs， providing essential technical support for efficient forward design.