Purpose The parameterization method commonly used in aerodynamic shape optimization has the characteristics of generalization, which usually leads to the need for a large number of design variables to achieve good optimization results in the inexperienced design process, and the design space cannot be flexibly defined according to the specific optimization problem. This study aims to propose an adaptive parametric method to solve this problem. Design/methodology/approach The adaptive parameterization method defines a rough initial design space, and then gradually refines the design space according to the sensitivity analysis method considering the optimization objectives and design constraints and the self-developed knot insertion technology, so as to maximize the improvement of the design space with fewer design variables during the optimization process. To evaluate the efficiency of this method, it is combined with the Kriging model with improvement expectation and genetic algorithm to construct a global optimization framework based on the adaptive parameterization method. Findings The drag reduction aerodynamic optimization design for RAE 2822 under transonic conditions is carried out. Compared with PARSEC and traditional fixed Hicks–Henne parameterization, adaptive parameterization has significantly improved the optimization effect at the same optimization cost. Further research shows that the optimization starting from NACA 0012 obtains a shape that is almost the same as the optimization starting from RAE 2822. Originality/value The adaptive parameterization method greatly reduces the dependence of the optimization results on both the initial shape and the design space.
Zheng et al. (Fri,) studied this question.