A new neural network-based optimal mixed H2/H1 control for a modified unmanned aerial vehicle subject to control input constraints

Unmanned aerial vehicles are subject to complex aerodynamic or hydrodynamic forces and various uncertainties, making their dynamic modeling difficult. Hence, in these systems, it is preferred to apply control techniques whose closed-loop systems can predict the system's uncertain dynamics and provide acceptable performance. To this end, this paper exploits a new neural network-based opti-mal mixed H2=H1 control for a modified unmanned aerial vehicle to accomplish trajectory tracking missions. Firstly, the dynamic model of the modified unmanned aerial vehicles is presented. Then, the design procedure of the new controller is delineated. In this approach, H1 attenuates the effect of uncertainties and through H2 the consumed control energy is minimized. Similar to real-world applications, it is supposed that there are control input constraints and external disturbances. The neural networks are also employed to estimate all uncertainties. The numerical simulations for two different desired paths are demonstrated. Simulation results illustrate the efficient per-formance of the proposed control technique in the presence of external disturbances, dynamic uncertainties, and control input constraints. (c) 2022 COSPAR. Published by Elsevier B.V. All rights reserved.