The correction of the thermal network model of space instrument based on the Kriging surrogate model

The thermal system of space instruments usually adopts a node thermal network model to determine the thermal design parameters. To facilitate the calculation, the thermal model based on the lumped parameter method will make a lot of simplifications and assumptions on the model parameters, resulting in deviations between the calculation results of the thermal model and the experimental temperature data. Traditional means of model optimization are costly and computationally slow, severely limiting the optimization space and efficiency. In this paper, an optimization framework combining a kriging surrogate model and a genetic algorithm is used to optimize the temperature response of four temperature sensors on a space instrument. The results show that the Root Mean Square Error (RMSE) between the optimized thermal network model temperature and the experimental temperature decreases from 1.99deg C to 1.23deg C, achieving significant optimization results.