Aiming at the requirement of aerodynamic parameters for the design of flight control system in the development of small unmanned aerial vehicles(UAV), an adaptive genetic algorithm combined with the maximum likelihood criterion is proposed to identify the aerodynamic parameters of the UAV. The algorithm introduces nonlinear adaptive values for ranking selection, adaptive mutation strategy and local hybridization, which can reduce the phenomenon of premature convergence and stagnation of traditional genetic algorithms. Using the program-controlled flight test data of a tandem wing UAV to test and verify the algorithm, the results show that the algorithm has good practicability and is applicable to all fixed-wing UAVs.