Thermal analysis model correction method based on Latin hypercube sampling and coordinate rotation method

Abstract

Abstract To predict on-orbit temperatures of the spacecraft accurately, a correction method for the thermal analysis model is proposed that combines Latin hypercube sampling and the coordinate rotation method. First, the thermal design parameters to be corrected are determined. Second, the thermal design parameters are sampled by the Latin hypercube sampling, and the sampling results inputted to the I-deas/TMG software for the thermal simulation calculation. Then according to the results of thermal simulation, the thermal design parameters are classified using the Spearman rank correlation coefficient formula, as either global sensitive parameters, local sensitive parameters, or insensitive parameters. Finally, according to the layered correction idea, the above parameters are corrected by the coordinate rotation method to find the optimal solution that satisfies the objective function. After the correction, the maximum temperature difference between the thermal simulation and the thermal balance test for each temperature measurement point is less than 1 °C for high-temperature conditions, and a maximum temperature difference of less than 2 °C for low-temperature condition.