Applicability analysis of genetic algorithms using in aerodynamic optimization

Genetic Algorithm(GA) has been applied in aerodynamic optimization for three reasons: First, it can theoretically find global optimum of complex optimization problems. Secondly, sensitivity calculation is not needed in the optimization process. Thirdly, it does not have any special requirement for design space. However, it is always very difficult to satisfy the theoretical convergence condition and severe problems on computation efficiency are confronted in practical application. It is considered that characteristics of aerodynamic problem raise a claim on algorithms parameter preference. In this thesis, the rule of GA parameter preference is discussed on the foundation of both algorithm process and aerodynamic problem analysis. Affect on airfoil shape optimization by variation of variable dimension in parametric model are analyzed and summarized. Elementary and quantitative rules for parameter preference are given. The results suggest that appropriate variable dimension should be set to describe the design space. Then population size and maximum generation should be set according to variable dimension. Some qualitative and partly quantitative results in this thesis can be used as basis for algorithm parameter preference on aerodynamic optimization design.