In this study, we performed a thermal simulation analysis of a space instrument, a solar spectrometer. A thermal model updating method was used to introduce the Kriging model as the surrogated model into optimizing thermal design parameters instead of directly iterating the finite element analysis. The sensitivity analysis method was used to eliminate the insensitive parameters, thus determining the influence area of modeling parameters and saving processing time. The valid parameters were then used in Latin Hypercube Sampling (LHS) to generate training samples. Eight Kriging models were constructed by the training samples, and a Genetic Algorithm (GA) was used to find the optimal set of parameters, under which the temperature values at certain positions of the model were closest to the results of the heat balance experiment, thus updating the thermal model. The proposed method was successfully performed on the thermal design of a space instrument. Using this model, temperatures of specialized positions predicted by the updated model were more precise than the initial ones with the RMSE of temperature deviation of 0.88 °C. The surrogate model updating technology based on Kriging is rapid and efficient for the iterative thermal design of aerospace products.

在这项研究中，我们对空间仪器、太阳光谱仪进行了热模拟分析。采用热模型更新方法，将克里金模型作为替代模型引入热设计参数优化中，而不是直接迭代有限元分析。采用敏感性分析方法剔除不敏感参数，从而确定建模参数的影响区域，节省处理时间。然后在拉丁超立方采样 (LHS) 中使用有效参数来生成训练样本。由训练样本构建8个克里金模型，利用遗传算法（GA）寻找最优参数集，使模型特定位置的温度值与热平衡实验结果最接近，从而更新热模型。所提出的方法已成功应用于空间仪器的热设计。使用该模型，更新模型预测的特定位置的温度比初始位置更精确，温度偏差的均方根误差为 0.88 °C。基于克里金法的代理模型更新技术对于航空航天产品的迭代热设计是快速高效的。