Real-Time Data-Driven Inverse Heat Conduction Method for a Reentry Flight Vehicle Based on the Random Forest Algorithm

The real-time monitoring of heat fluxes on the surface of a flight vehicle is vital for safe reentry and efficient attitude control. However, conducting measurements during flight is a challenge. In this study, a real-time heat flux estimation method using inverse heat conduction was proposed, and internally mounted sensors were used for measurements. First, the time-varying samples of heat flux on the surface and temperature on the inner wall were generated along a variety of reentry trajectories. Second, a sensor selection algorithm based on feature importance was applied to select optimal sensor mounting locations. Finally, a prediction model was built using the random forest algorithm to estimate surface heat flux with measured temperatures from the mounted sensors. The proposed method was employed to predict the real-time heat flux of a reentry capsule during return flight. The results show that the proposed method can predict heat flux in real time with a prediction error of less than 0.2%. Further, the sensor selection algorithm enhanced prediction efficiency by reducing the number of necessary sensors.