Research On Nonlinear Filtering Technology And Its Application In Deep Space Exploration Autonomous Navigation

With the official establishment of the national Mars exploration mission,China is making great strides forward in its Mars exploration.Mars is between 50 and 400 million kilometers away from Earth.Such a long distance means a huge communication delay,which causes the detector must be able to autonomously correct all kinds of errors during the cruise to complete the arduous task,and the filtering algorithm plays a key role in correcting the deviation of navigation position and speed.The deep-space environment is complex and changeable,which leads to the nonlinearity of the orbital dynamics model and the navigation system model of the detector,and only the nonlinear filtering algorithm can be used to process these above models.Therefore,this thesis focuses on the nonlinear filtering algorithm,especially particle filtering,and the goal is to improve particle filtering.Simultaneously,the application scenario of this thesis is the Mars exploration cruise segment.In order to apply the nonlinear filter algorithm to the process of detector cruise,it is necessary to establish an effective detector autonomous navigation system.The following are the main works of this thesis.1.According to the navigation requirements of the detector cruise segment,this thesis analyzes the advantages and disadvantages of various navigation methods,and finally selects optical autonomous navigation as the autonomous navigation method of this thesis.Under the simulation conditions,the most important four steps to establish an optical autonomous navigation system are: establishing the state equation and measurement equation of the system,establishing the nominal orbit of the cruise segment detector,selecting the navigation landmark and completing the planning of the navigation landmark,and the filtering algorithm.In this thesis,the state and measurement equations of the navigation system are established by studying the orbit dynamics and optical camera imaging.The nominal orbit is established by using STK instead of MATLAB.At the same time,after careful analysis,the asteroid is determined as the navigation roadmap of this thesis.On the basis of the existing asteroid planning algorithm,the principle of optical autonomous navigation based on asteroid navigation is studied in depth,and this thesis proposes an asteroid programming algorithm with simple operation.We can find that the screening algorithm is simple and effective through simulation experiments.2.In view of the complex and variable environment of deep space environment,the nonlinear filtering algorithm is studied,especially particle filtering.Aiming at the particle degradation problem,an unscented particle filtering algorithm based on weight selection and weight optimization is proposed.In the process of particle importance sampling,unscented Kalman Filter is introduced to form unscented particle filtering.At the same time,in the process of particle resampling,perform weight selection and weight optimization,so that small weight particles are not easily discarded,ensuring particle diversity.Through the improvement of the above two aspects,the particle degradation problem is solved to some extent,and the filtering precision is improved.3.In the context of the 973 deep space exploration project,the simulation of the autonomous navigation system is established.The simulation experiments show that the improved particle filter has higher filtering accuracy than PF and extended Kalman filter.