Research On The Attitude And Navigation Information Fusion Algorithm Of Multi-rotor UAV

Multi-rotor UAV has been widely used in the military and the civil fields.Navigation system, which is an important part of multi-rotor UAV, is the base of the security and the stability in flight. The INS/GPS integrated navigation system used can realize a high precision navigation. The integrated navigation system has the characteristics of the complementary advantages and the redundancy of navigation mechanism, the essence of which is a processing system for the multi-sensor navigation information. The main navigation parameters of UAV can be obtained by means of the multi-sensor information fusion. So the information fusion technology is the key technology of the integrated navigation system, and has become a research hotspot of scholars at home and abroad. In this paper the new and small UAV of a multi-rotor structure developed by our group is used as a research platform, to launch the research on the key technology of the information fusion of an airborne multi-sensor integrated navigation system. The thesis includes the following aspects:(1) The measurement application characteristics of the airborne sensors on the multi-rotor UAV are researched, focusing on the noise source of a gyroscope and the method of eliminating the error. On this basis, it is determined the multistage information fusion structure of the attitude, position and speed, which is based on a hierarchical structure design of the information fusion, can significantly reduce theamount of calculation of the navigation system, and can improve the maneuvering characteristics of the system. The every level of multistage fusion structure can adopt different fusion algorithms, focuses on the Kalman filter algorithm in multi-sensor integrated navigation system, in which the application is the most successful and the most extensive.(2) According to the analysis of the measurement characteristics of each sensor,the paper does some research about the algorithm of calculating navigation information by each sensor. A local geographical navigation coordinate system is used in a mechanical layout, and the corresponding simplified is made according to the actual UAV flight characteristics. On this basis, we research the attitude calculation algorithm by the gyro, and by the accelerometer and the magnetometer,the position and the velocity calculation algorithm by the accelerometer and by the GPS. The chapter lays the foundation for establishing the information fusion model and algorithm.(3) The first level attitude information fusion is carried on by using the gyro, the accelerometer and the magnetometer, using Kalman filter algorithm for the fusion algorithm. The nonlinear discrete-time state space model is established for the attitude fusion system, using the extended Kalman filter to solve the linearized problem of the model. The paper presents an improved Sage-Husa adaptive extended Kalman filter algorithm, in which the attitude angle variance of the gyro dynamic calculation is used to estimate the system noise variance, using the adaptive filtering algorithm for the online real-time estimation of the measurement noise variance, it can guarantee the precision and the stability of filtering. The algorithm is also introduced into the filter convergence criterion, combined with the strong tracking Kalman filter algorithm effectively suppresses the filtering divergence problem.(4) The second level of the horizontal position and the velocity information fusion is carried on by using the accelerometer and the GPS, using the improved algorithm of Kalman filter, establishing a linear discrete-time state space model. In this model, the acceleration information as the control input of the state equation topredict the position and the velocity information indirectly, which is conducive to improve the accuracy of information fusion. The barometric altimeter is introduced to achieve the information fusion of the third level vertical height and velocity combined with the accelerometer and GPS. According to the design idea of federated filter, it is determined two times fusion structure, respectively, using the improved Kalman filter algorithm and the weighted least squares estimation algorithm. Before the second time fusion, the GPS fault diagnosis is increased to enhance the fault-tolerant ability of the information fusion system.At the end of the thesis, the work of the full text is summarized, and the future work is prospected.