Aircraft optimization design problems are mostly computationally intensive. These complicated problems probably contain mixed-variables, while most research has focused on continuous variables. This article sets up a mixed-variable surrogate-based optimization algorithm framework that includes a mixed-variable experiment design method and an inaccurate infilling method. The mixed-variable experiment design method combines improved successive local enumeration with enhanced stochastic evolution to deal directly with discrete variables. The inaccurate infilling method tends to find points with better fitness value and relatively low sample density so as to balance exploration and exploitation. Several numerical functions and a mixed-variable solid rocket motor performance matching design problem are solved using the modified surrogate-based optimization method. The results indicate that the proposed method is competitive compared with other heuristic algorithms and surrogate-based algorithms, and can deal with mixed-variable aircraft design problems effectively.

飞机优化设计问题大多是计算密集型的。这些复杂的问题可能包含混合变量，而大多数研究都集中在连续变量上。本文建立了一个基于混合变量代理的优化算法框架，包括混合变量实验设计方法和不准确填充方法。混合变量实验设计方法将改进的连续局部枚举与增强的随机演化相结合，直接处理离散变量。不准确的填充方法倾向于找到具有较好适应值和相对较低样本密度的点，以平衡探索和开发。使用改进的基于代理的优化方法解决了几个数值函数和混合变量固体火箭发动机性能匹配设计问题。结果表明，与其他启发式算法和基于代理的算法相比，所提出的方法具有竞争力，可以有效地处理混合变量飞机设计问题。