Attitude Control Of Quadrotor UAV Based On Neural Network

In recent years,the quadrotor unmanned aerial vehicles(UAV)have become the beach-goers of the times.Because of their such advances,such as low cost,no casualties,good motor performance and simple mechanical structure,they play a very important role in military and civilian fields,for example,they are applied in enemy investigation,disaster detection and traffic management.The quadrotor UAV's control system is highly non-linear and underactuated,which makes its control problem more complex and more and more people are committed to studying its control technology.The main task of this paper is to design the controller based on the neural network for the attitude control system of the quadrotor UAV,and complete the tracking control of its attitude.The attitude control system is a multi-input and multi-output nonlinear system,which can be regarded as a system composed of three strict-feedback subsystems.As the discrete-time system is more close to the actual controlled object,firstly,we transform the attitude control system into discrete-time system,then by defining the coordinate transformations,the discrete-time system is transformed into a special form,which is suitable for the backstepping design of the tracking controller.The neural network controller designed by this method avoids the possibility of controller singularity,and also avoids the noncausal problem when the discrete system uses the backstepping method to design the controller.In this paper,the high-order neural networks(HONN)are utilized to approximate the unknown virtual controllers and the actual controllers,and we needn't to know any priori knowledge of them.The Lyapunov function method proves that under certain relaxed assumptions,the neural network controller designed in this paper can gurantee that the closed-loop system is stable in sense that semiglobally uniformly bounded of all the signals and the tracking errors converge to a bounded compact set.In order to verify the correctness and validity of the designed attitude controller,this paper uses Matlab software to simulate the designed controller.The simulation results show that the controllers and the adaptation laws can achieve our expected target,what's more,the neural controllers have an certain anti-interference ability.