The high-pressure turbine blade in aero-engine power system may experience microstructural degradation due to uncertain flaws, multi-physical fields and loads during manufacturing, processing, installation, and maintenance, leading to serious structure deterioration that affects safety and reliability in service. Therefore, it is necessary to assess the influence of random flaws and loads on the fatigue performance of turbine blades from a probabilistic perspective. In this study, we propose a novel method based on the Artificial Hummingbird Algorithm and Kriging surrogate model (AHA-Kriging), for flaw tolerance assessment in the surface partition of the turbine blade. The results indicate that in the hazardous zone, the flaw tolerance reliability is 0.9984, corresponding to a LCF life of 1520 cycles. In the safe zone, the flaw tolerance reliability is 0.9991, corresponding to a LCF life of 2501 cycles. The primary factor influencing LCF life is flaw size, followed by factors such as the strength coefficient, gas temperature, and fatigue strength exponent. Besides, the AHA-Kriging approach exhibits higher modeling precision and simulation efficiency compared to other methods. This paper presents a practical engineering approach for assessing flaw tolerance in the surface partition of complex components, which is of significant value.