Integration of CNN in a Dynamic Model-Based Controller for Control of a 2DOF Helicopter With Tail Rotor Perturbations

This paper shows a proposal for control scheme the trajectory tracking problem in Two Degree of Freedom Helicopter (2DOFH). For this purpose, based on feedback linearization combined with Generalized Proportional Integral (GPI) controller is used. In order to implement by feedback, it required know and have access all physical 2DOFH parameters, however, angular velocity viscous friction are often not available. Commonly, state observers used velocity, estimating results out be more complex. Therefore, we propose use Convolutional Neural Network (CNN) estimate velocity. The variables estimated CNN entered into both GPI feedforward controllers. Thus, system brought linear representation that directly relates dynamics perturbations non-model parameters. Finally, numerical simulations shown validate robustness our presence disturbances tail rotor, as well advantages using CNN.