Aimed to gain a high aerodynamic efficiency in various flight conditions, the aerodynamic optimization of the adaptive airfoil leading edge was conducted based on Isight platform. First, a modified method of original Hicks-Henne shape function was adopted to parametrically describe the leading edge. Then, Latin hypercube sampling was applied to generate the samples, whose aerodynamic performances were obtained from a series of CFD simulations. Subsequently, those simulated data sets were used to train the RBF neural network. Finally, the optimization of the objective function using the multiisland genetic algorithm(MIGA) based on the neural network approximation model was performed to achieve the optimal results. The lift to drag ratio optimization design of NACA 0006 was performed as an example using this combined strategy. The results show that the modified Hicks-Henne shape function describes the airfoil leading edge with satisfaction. The combined optimization method evidently improves the efficiency of the airfoil aerodynamic optimization procedure.