Research And Application Of Memetic Algorithm In Wing Design

Traditional optimization methods have rapid convergence quality and high local search capability, since they can take full advantage of problem information and the"neighborhood"to find the next solution from an initial solution in search space according to a certain principle.Genetic algorithms (GAs) are population-based parameter search procedures simulating evolution process in environment. GAs are capable of exploring and exploiting promising regions of the search space for they search from a group of initial solutions. But because of strong randomicity, GAs suffer from low local search capability and poor convergence quality.To solve the efficiency problems of GAs, several traditional Memetic Algorithms (MAs) ,hybrid between GAs and several local search methods, are presented and tested in this paper.The choice of local search methods or memes significantly affects the performances of searches. To solve this problem, we present a Diversity-based Adaptive Memetic Algorithm (DAMA), which can select specific local search method for every individual adaptively as search progresses. Experimental studies on continuous parametric benchmark problems show that DAMA have stronger efficiency and universal than traditional MAs. To continually increase efficiency, a parallel DAMA is achieved in matlab environment.At last, the algorithms proposed are applied to airfoil and wing optimization, and encouraging results have been obtained.