Data-driven model for predicting surface heat flux of hypersonic vehicles based on eXtreme Gradient Boosting

To efficiently and accurately obtain hypersonic vehicle surface heat flux data and shorten the design cycle of thermal protection systems. In this study, a data-driven model is developed to predict surface heat flux on hypersonic vehicles, taking into account the influence of complex surface geometrical features (SGF), a data-driven model, based on eXtreme Gradient Boosting (XGBoost), called XGBoost_SGF. First, the flight state parameters (Height, Mach number, Ma∞, and angle of attack, α) and the state parameters of physical characteristics (atmosphere density, ρ∞, static temperature, T∞, total temperature, T0, and static pressure, P∞) are constructed. Second, surface geometric features are extracted from the arithmetic combinations of eigenvalues of the covariance matrix within the local neighborhood of the three-dimensional (3D) point cloud. Finally, the XGBoost_SGF model is trained using heat flux values corresponding to different spatial coordinate points as output targets. Validation on hypersonic double-ellipsoid configurations demonstrates that the XGBoost_SGF model achieves high predictive accuracy, outperforming versions that omit surface geometric features. The prediction error for the overall surface heat flux remains below 9%, while the error at the stagnation point is less than 2.5%. Moreover, the XGBoost_SGF model significantly enhances computational efficiency, requiring approximately 2 s to predict the surface heat flux distribution for a single flight condition. The data-driven XGBoost_SGF enables accurate and timely aerodynamic thermal analysis, effectively reducing the design cycle of thermal protection systems.