Application of adaptive differential evolutionary algorithms to aerodynamic optimization design

Based on real-coded differential evolutionary (DE) algorithm, a population entropy estimating method is introduced into DE to analyze the distribution of individuals in the population. Search range in this algorithm is adaptively changed according to individual distribution. Function optimization test suggests that this algorithm has better convergence stability and convergence speed. The Navier-Stokes equations are adopted for calculating the airfoil aerodynamic performances. Three optimization algorithms, i.e., the standard genetic algorithm(SGA), the DE algorithm, and the adaptive range differential evolutionary (ARDE) algorithm, are tested on an airfoil shape aerodynamic optimization problem. Comparison of the optimization results from the three different algorithms demonstrates that the ARDE algorithm has better search capability. According to the characteristics of the aerodynamic optimization problem, influences from the control parameters in the algorithm on optimization results are analyzed and studied.