Research On Attitude Control System Of Micro Quadrotor Based On MEMS Motion Sensor

Micro quadrotor is a kind of micro air vehicle which is based on the development of MEMS sensor in recent years.It has the characteristics of small volume,light weight,simple structure,flexible movement and has been widely used in civil and military fields.From the point of the system structure,the micro quadrotor is a nonlinear system with underactuation and strong coupling.Its motion process is closely related to the flying attitude in the meanwhile.Therefore,how to achieve the precise control of attitude angle has become the focus of this research.Firstly,the related studies and results of the micro quadrotor at home and abroad are described and summarized in this thesis.On this basis,the motion mode of the micro quadrotor is analyzed,and the corresponding dynamic model is established.Secondly,a new control algorithm based on PID neural network is proposed to solve the problem that traditional PID controller has difficulty in parameter setting and lack of control stability for strongly coupled system such as micro quadrotor.This algorithm integrates the PID control principle into the neural network and optimizes the control process by using the nonlinear mapping and infinite approximation of the neural network.In order to further improve the output response of PID neural network,a dynamic adjusting scheme for the proportional coefficient of state function in PID neuron is designed.The simulation results show that the modified method can accelerate the response speed of the system 0.472 seconds and the overshoot is only 0.76 % at the same time.Ultimately,the precise control of attitude angle is realized.Finally,in order to verify the validity of the algorithm,a microquadrotor based on STM32 controller is designed.Besides,the data fusion is accomplished by MEMS motion sensor MPU9250 and Kalman filter.In the end,motion control of the micro quadrotor is realized.The simulation by MATLAB and actual flight test show that compared with traditional PID controller,PID neural network has faster stable time and smaller overshoot,in the mean time generalized decoupling is realized as well,which improves the stability and adaptive ability of the system.