Real-Time Reconstruction of the Temperature Field of NSRT's Back-Up Structure Based on Improved RIME-XGBoost.

Obtaining an antenna's back-up structure (BUS) temperature field is an essential prerequisite for analyzing its thermal deformation. Thermodynamic simulation can obtain the structure's thermal distribution, but it has low computational accuracy. There is a problem with cumbersome wiring and difficult maintenance of the temperature measurement system. This study developed an improved RIME-XGBoost model to realize the temperature prediction of the BUS of the Nanshan 26-m Radio Telescope (NSRT). The proposed model successfully predicts the NSRT's BUS temperature distribution based solely on environmental sensing (ambient temperature, angle of solar radiation, antenna's orientation, etc.). The relative prediction accuracy between the predicted and actual BUS temperature is 97.15%, and the predictive error is less than 0.897 K (root mean square error, RMSE). This research result provides an alternative method for the real-time reconstruction of the structure's thermal distribution in large-aperture radio telescopes.