Dynamic Surface Neuro-Backstepping Based Flight Control With Asymmetric Output Constraints

In this article, flight path angle of the aircraft is controlled asymptotically for a class of nonstrict feedback system by employing robust backstepping dynamic surface control. First-order low-pass filters are used at each step of the backstepping architecture to implement constraints on the virtual control states, eliminating the explosion of complexity problems. A precise understanding of nonlinear aerodynamic forces and moments is vital to arrive at such a control approach under dynamic atmospheric circumstances. Adaptive radial basis function neural network (RBFNN) is used as an activation algorithm for the complex nonlinear function approximation by tuning the weights. This structure of RBFNN is employed in the backstepping framework which is embedded with the designed surfaces. Apart from the well-known physical properties, the formulation of such a law requires very little knowledge about the aerodynamic model. Asymmetric time-varying barrier Lyapunov function candidate is used to constrain the output from designed limit. In the notion of Lyapunov, closed-loop signals are theoretically demonstrated to be semiglobally uniformly ultimately bounded. Finally, the efficacy of the proposed law is tested in a number of simulation scenarios. The current study's findings indicate satisfactory control performance, with output tracking desired signals and system states remaining on the designed sliding surfaces.