In view of the effects of aerodynamic coupling and disturbance on unmanned aerial vehicles formation flight
process, a design method of neural network adaptive inversion based on “leader-wing” mode is proposed. Firstly, considering
the kinematics equations of the formation, the 3-D(three-dimensional) nonlinear mathematical model of the formation flight is
established. The basic control law is developed in nonlinear dynamic inversion. Then the nonlinear dynamic inversion errors
due to modeling error or disturbance are compensated adaptively on line by improved BP neural network. The performance
of the control system is improved. A simple and effective design idea for formation change is proposed. Simulations
demonstrate that the controller is effective and able to keep or change formation configuration rapidly, stably and exactly
with no collision, and has a good anti-interference performance.