Research On Multi-sensor Integrated Navigation System Of Small Unmanned Aerial Vehicles Based On Federated Filtering

Navigation and positioning technology is the key for unmanned aerial vehicles(UAVs)performing tasks.As the mission becomes more complicated and the flight environment becomes more diverse,as well as the defects and limitations of airborne sensors,the single navigation method cannot meet the daily application requirements.Making full use of the limited airborne navigation sensors to design a UAVs integrated navigation system that can meet the requirements of high precision,high fault tolerance and low cost,has important research meaning and broad prospects for development.In this thesis,the integrated navigation system suitable for small UAVs is taken as the research object,and the research on multi-sensor information fusion technology and integrated navigation system design is carried out.The main contents are as follows:First of all,the working principle and error sources of the most widely used strapdown inertial navigation system(SINS)and global positioning system(GPS)are explored and analyzed.The mathematical model of SINS is established,and the sources of SINS errors are analyzed.Based on the low-precision and low-cost MEMS(Micro-Electro-Mechanical Systems)inertial devices commonly used in small UAV,the modeling and quantitative analysis work of errors are carried out by combining the actual test and Allan variance method.The principle and error sources of GPS navigation are described in detail.The errors of GPS single point positioning are calculated through outdoor comparative experiments.It is verified that the accuracy of GPS single point positioning is easily affected by its own hardware conditions,the surrounding satellite signal environment and other factors.Then,the federated filtering integrated navigation technology for small UAVs is studied.An integrated navigation structure is designed using federated Kalman filter based on the small UAV airborne sensors as the information sources,and the mathematical model of multi-sensor information fusion is established.Two new measures of offset factor and oscillation factor are proposed to evaluate the performance of local filter and to dynamically adjust the information distribution factor.On the basis of this,a matrix-based adaptive information distribution algorithm is designed to realize the information distribution and integration of each estimated state independently.Thirdly,the simulation results show that the proposed adaptive information allocating algorithm has advantages compared with the traditional methods.The integrated navigation simulation system is established and many different simulation conditions are set up.The simulation shows that this method can adaptively distribute the information according to the filter performance among all local filters,and can reduce the fusion weight of the problem filter in time,so as to avoid polluting the whole filtering structure and to reduce the influence of external interference.Finally,the practical applicability and effectiveness of the proposed adaptive algorithm are verified through outdoor UAV flight experiments.The self-developed six-rotor UAV is used to implement outdoor static experiments and dynamic flight experiments.Both the proposed adaptive algorithm and the traditional federated integrated navigation algorithm are used to compute the navigation information.Experimental results show that the adaptive algorithm can effectively improve the accuracy of the integrated navigation system in practical applications.