In high thermal-load turbines, the cooling layout design for hot-section components is typically based on deterministic design for specific operating conditions[1]. However, neglecting the inherent uncertainties associated with manufacturing variability and operational conditions can result in pronounced discrepancies between theoretical cooling performance and its realistic behavior. There is still a significant lack of robustness structural design approaches for turbine hot-section components[2], leading to severe material erosion dangers when facing uncertainty factors. Therefore, this study introduces an efficient robust optimization method which bridges the optimization and uncertainty spaces by mapping sample information between them and establishing an information channel through the expected improvement mechanism, thereby optimizing the utilization of available sample data obtained by costly numerical calculations. The proposed algorithm is applied to the robust optimization design of a gas turbine partial squealer tip. Wind tunnel experiments were designed and conducted on the blade tip for validating the robustness of the optimization design. Corresponding numerical simulations are also conducted to analyze the robustness mechanisms of aero-thermal performance of the robustness optimization design by comparing with sample data. During robustness optimization, the proposed algorithm reduces the sample requirement by 38.46%. The experimental results showed that the optimized structure reduces the blade tip wall temperature by 5.13K, while the aerothermal robustness when varying operating conditions is significantly enhanced compared with the baseline design, especially for the middle and downstream tip regions. Further flow thermodynamic analyses show that the optimal structure increases the proportion of coolant jet that is suppressed by the rim gap leakage flow and remains in the cavity, which reduces the interaction between the coolant jet and unsteady leakage vortices, thereby improving stability and minimizing variability in performance. This paper provides technical support for the high-efficiency robust design method of turbine blade tip cooling structures.