Highly Elliptical Orbit Satellite Autonomous Navigation Based On Adaptive UKF And Position/velocity Measurement Assistance

In highly elliptical orbits, the satellite orbit dynamics change severely and willexperience a variety of harsh space environment. Therefore, two new issues areraised to highly elliptical orbit satellite autonomous navigation system:1) The lownavigation precision caused by the uncertain measurement noise from the harshspace environment;2) The navigation estimate accuracy is fluctuate due to highlyelliptical orbit dynamic characteristics. To solve the problems, this paper carried outthe follow study in order to achieve highly elliptical orbit satellite autonomousnavigation.Strong tracking adaptive UKF filter is studied. As the standard UKF algorithmis invalid to uncertain measurement noise, promote the idea of strong trackingadaptive filter to UKF filter. According to orthogonality principle, the strongtracking adaptive UKF is deduced through introducing adaptive factor. Thisalgorithm inherits the adaptive ability to uncertain noise of strong tracking adaptivefilter, omittes the calculation of Jacobin and achieves three order precision.As the calculation of UKF is increased due to adaptive computing andUnscented Transformation, further methods to cut down calculation and improvecomputational efficiency are studied. If the uncertain measurement noise exists onlya short period of time, the filter need not perform adaptive calculation, but startadaptive calculation only when uncertain noise is detected. Based on this idea,switch adaptive UKF is studied. Through the switching law based on large numberstheorem, the filter mode is judged and switched. The Unscented Transformation isthe key point of UKF algorithm. The estimate precision is improved using Sigmapoint directly which approaches the probability density of nonlinear functioninstead of linearization. However, the dimension of system through UnscentedTransformation is2n+1which increased calculation. Therefore, the hypersphericaltransformation method instead of Unscented Transformation is studied. Thedimension of Sigma point is reduced to n+2using this method and the calculation iscut down significantly. Compared with switch adaptive UKF, this method reducedcalculation absolutely.Switch fuzzy adaptive UKF is studied. The adaptive factor is further regulatedand the adaptive estimate precision is improved further more by fuzzy logic. Inorder to improve the adaptive ability of fuzzy adaptive UKF to overcome high leveluncertain noise, fuzzy membership functions are established in two modes: thenormal filter mode and the filter mode under high level noise. The purpose ofadaptive filter in normal mode is to get faster convergence and overcome theuncertain noise in the mode under high level noise. In order to deal with the disadvantages of learning ability and fuzzy inference, BP Neural Networks isintroduced to the filter process. The Neural Networks is connected with the fuzzysystem in series in order to improve the fuzzy adaptive factor.Autonomous navigation methods suitable for highly elliptical orbit satellite arestudied. In order to overcome the filtering problem at perigee using celestialnavigation, the position/velocity measurements auxiliary correction method isintroduced. Two proposals are maily studied: autonomous navigation method basedon radar altimeter measurement and the auxiliary correction method using GPS. Inthe first proposal, the radar altimeter measurement is used as supplement tostar-earth angular observations. The radar altimeter application range andmeasurement principle is presented, the measurement equation is established, andthen the autonomous navigation method based on celestial&radar altimeter isproposed. In the second proposal, the coverage of GPS to highly elliptical orbitsatellite is analysised. According to the GPS position/velocity measurements, theGPS measurement equation is established and applied to navigation filter. Throughinformation fusion with celestial navigation, the high navigation output is achieved.Finally, autonomous navigation experiment system is established, the strongtracking adaptive UKF algorithm and the switch fuzzy adaptive UKF algorithm areapplied to highly elliptical orbit satellite autonomous navigation experiment. Theexperiment is performed under different uncertain measurement noise levels. Theresults of the experiment show that the autonomous navigation methods put forwardby this thesis is effective and feasible.