The Nonlinear Filter Based On The Measurement Information And Its Application Research In Navigation System

Deep space detection is one of the world’s main hotspot in the space activities in the new century, it’s technology progress gradually, which has made remarkable progress in navigation technology of deep space probes, and provides the technical foundation for a series of deep space missions, such as unmanned detection and manned detection. Effective navigation and control is one essential prerequisite for the success of deep space exploration missions. The implementation of deep space probes autonomous management can reduce the cost of space programs, and can guarantee deep space probes have good independent ability to survive. Filtering methods play a key role in the process of navigation, when the observation model and system model of filter have serious nonlinearity, the navigation system can only use filtering methods which can be applied in nonlinear systems, if not the filtering performance is likely to be disappointing. The navigation system adopted in this thesis happens to this type, and some entries be regarded as disturbance in system model can are hard to be regarded as Gaussian noise, for this reason, Unscented Particle Filter is the most suitable filtering method.This thesis combines the national key basic research development plan project ‘the basic research of deep space detection high precision of astronomical angular and velocity integrated autonomous navigation’, according to the characteristics of the Mars cruise period, do some further study in relevant autonomous navigation methods, and on this basis, improve the nonlinear filtering method based on measurement information. The main content of this thesis includes the following several aspects:On the one hand, this thesis takes navigation methods into account, it analyzes the particularity of deep space detection environment, study the observability analysis methods of autonomous navigation systems’ navigation celestial bodies in such an environment, and through such observability analysis method to analysis the observability of autonomous navigation systems, research autonomous navigation methods based on asteroids’ information of cruise period. This thesis utilizes information fusion method to combine celestial navigation and redshift navigation, and then build a Celestial Navigation System/Redshift Navigation System(CNS/Redshift) integrated navigation system, using angular information and velocity information at the same time, so it can get higher navigation accuracy.On the other hand, this thesis takes filtering algorithm into account, it studies the traditional unscented particle filter algorithm, and it’s improved algorithm. This thesis proposed two kinds of improved unscented particle filters, both of the two utilize spherical simplex unscented transform to replace the traditional symmetric sets unscented transform, thus can significantly reduce the number of sampling points in filter process, and then the computational burden can be reduced. In the first improved unscented particle filter, iterative strategy is adopt, it utilizes new observation information to linear points, and then improve the navigation filter accuracy by iterative update. The second improved unscented particle filter algorithm adopts adaptive method which based on the innovation, makes filter have the ability of adaptive, and its accuracy has a certain amount of improvement.After that, the thesis apply improved unscented particle filer to CNS/Redshift integrated navigation system, and computer simulation in MATLAB, analysis the simulation results, we can found compared with traditional unscented particle filter, improved algorithms have better navigation performance.