传统的气动计算方法计算繁琐、计算效率低,不适应于乘波体多学科设计优化,通过建立气动代理模型可以很好地解决气动计算精度和效率的矛盾。利用面元法进行气动估算,采集了锥导乘波体在设计点、非设计点的气动特性作为训练数据,构建了Kriging和LS-SVM代理模型,对比了两种模型对此高维问题的代理效果。结果表明,Kriging代理模型能更准确地表达锥导乘波体的气动特性,应用代理模型进行优化等工作的计算效率与传统气动计算方法相比有显著的提高。 The traditional calculation method of pneumatic calculation is cumbersome, the computational efficiency is low, which is not suitable for multidisciplinary design optimization by multiplication. By establishing the aerodynamic agent model, the contradiction between accuracy and efficiency of aerodynamic calculation can be well solved. The aerodynamic characteristics of cone-guided multiplication wave body at design point and non-design point are collected as the training data by aerodynamic estimation using the face-element method. The Kriging and LS-SVM agent models are constructed and compared with the two models for the high-dimensional problem Agent effect. The results show that the Kriging agent model can express the aerodynamic characteristics of the cone-guided multiplicator more accurately, and the computational efficiency of the agent-based model is significantly improved compared with the traditional aerodynamic calculation method.