Research on Unsteady Aerodynamic Modeling Method Based on Deep Neural Network

The innovative design of the typical nonlinear unsteady aerodynamics of the aircraft during flight is the research focus of researchers in related fields in the industry. The algorithm model based on deep learning created for the high angle of attack range of innovative design aircraft is of great research value for aircraft flight simulation calculation and flight control law design. Aiming at the high-maneuverability aircraft with complex motion, the aerodynamic prediction model based on bidirectional-long short-term memory neural network is established by using the real data of large pitch oscillation obtained from wind tunnel test. Then, the transfer learning method is used to fine-tune the parameters and enhance the performance of the deep learning model. Finally, the effectiveness and generalization of the algorithm model are verified by the typical maneuver simulation actions' tail dash ‘ and ‘ cobra ‘ test in wind tunnel test. The results show that the deep learning algorithm adopted in this paper can accurately predict the unsteady aerodynamic force of the standard model aircraft with small aspect ratio flying wing layout under different maneuvers, and the algorithm model enhanced by transfer learning technology has a good performance improvement on the accuracy of the algorithm.