Genetic algorithm LQG and neural network controllers for gust response alleviation of flying wing unmanned aerial vehicles

In this paper, a genetic algorithm linear quadratic Gaussian controller (GA-LQG) and an artificial neural network (ANN) controller are implemented for gust response alleviation of lightweight flying wings undergoing body-freedom oscillations. A state–space aeroelastic model has been formulated by coupling the unsteady vortex lattice method for aerodynamics with finite-element based structural dynamics. The model is subsequently reduced using balanced truncation to improve computational efficiency during controller synthesis. Open-loop simulations show that the flying wing experiences large changes in pitching angles during gusts. For GA-LQG controller, the LQG weights are optimised using a genetic algorithm, maximising a defined fitness function. Generally, the GA-LQG controller reduces the plunge displacements by up to 94.2% while damping out wingtip displacements for discrete and continuous gusts. Similarly, the ANN controller effectively regulates both the plunge displacements and wingtip displacements, including gust cases that are not presented during the ANN training phase. The ANN controller is more effective in correcting wingtip displacements during discrete gusts than the GA-LQG controller, while the opposite is true for the continuous gust cases. The ANN controller offers several advantages over the GA-LQG controller, including the elimination of the need for a Kalman filter for full state estimation and offers a non-linear control solution.