Fast evaluation of film cooling effectiveness under radiative effects with attention mechanism

In the design of cooling structures for aero-engine turbine blades, film cooling effectiveness serves as a primary validation parameter, and its rapid assessment is essential to achieving accelerated blade design. Furthermore, the comprehensive inclusion of radiation effects in the film cooling design process is crucial for advancing toward refined turbine blade design. Therefore, the rapid evaluation of film cooling performance under radiative influence is of significant research importance. This study employs neural networks and incorporates radiative effects to develop a fast-prediction model for film cooling effectiveness, using key operating parameters and wall curvature as inputs. A comparative analysis was also conducted to evaluate the enhancement effects of different attention mechanisms on the predictive network. The results indicated that the SimAM mechanism outperforms the combination of the spatial attention mechanism and the channel attention mechanism in terms of prediction accuracy and training efficiency. Tests demonstrated that the proposed film cooling prediction network achieves high accuracy, with a mean absolute prediction error below 0.004. The network successfully captures the complex nonlinear effects of key operating parameters and model curvature on film cooling effectiveness, indicating that under the influence of radiation, the cooling performance of convex models is superior to that of flat and concave models. These findings can contribute to the rapid and refined design of turbine blades.